Datasets:

---

bigcode

/

jupyter-parsed

like 4

Follow

BigCode 1.4k

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "!pip install plotly", "_____no_output_____" ] ], [ [ "<a href=\"https://plotly.com/python/\" target=\"_blank\">Plotly's</a> Python graphing library makes interactive, publication-quality graphs. Examples of how to make line plots, scatter plots, area charts, bar charts, error bars, box plots, histograms, heatmaps, subplots, multiple-axes, polar charts, and bubble charts.", "_____no_output_____" ] ], [ [ "!pip install plotly_express", "_____no_output_____" ] ], [ [ "<a href=\"https://pypi.org/project/plotly-express/0.1.9/\" target=\"_blank\">Plotly Express</a> is a terse, consistent, high-level wrapper around Plotly.py for rapid data exploration and figure generation.", "_____no_output_____" ] ], [ [ "!pip install calmap", "_____no_output_____" ] ], [ [ "<a href=\"https://pypi.org/project/calmap/\" target=\"_blank\">Calendar heatmaps (calmap)</a> Plot Pandas time series data sampled by day in a heatmap per calendar year, similar to GitHub’s contributions plot, using matplotlib.", "_____no_output_____" ] ], [ [ "!pip install squarify", "_____no_output_____" ] ], [ [ "Pure Python implementation of the <a href=\"https://pypi.org/project/squarify/0.1/\" target=\"_blank\">squarify</a> treemap layout algorithm. Based on algorithm from Bruls, Huizing, van Wijk, \"Squarified Treemaps\", but implements it differently.", "_____no_output_____" ] ], [ [ "!pip install pycountry_convert", "_____no_output_____" ] ], [ [ "Using country data derived from wikipedia, <a href=\"https://pypi.org/project/pycountry-convert/\" target=\"_blank\">pycountry-convert</a> provides conversion functions between ISO country names, country-codes, and continent names.", "_____no_output_____" ] ], [ [ "!pip install GoogleMaps", "_____no_output_____" ] ], [ [ "Use Python? Want to geocode something? Looking for directions? Maybe matrices of directions? This library brings the <a href=\"https://pypi.org/project/googlemaps/\" target=\"_blank\">Google Maps</a> Platform Web Services to your Python application.", "_____no_output_____" ] ], [ [ "!pip install xgboost", "_____no_output_____" ] ], [ [ "<a href=\"https://xgboost.readthedocs.io/en/latest/\" target=\"_blank\">XGBoost</a> is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable. It implements machine learning algorithms under the Gradient Boosting framework. XGBoost provides a parallel tree boosting (also known as GBDT, GBM) that solve many data science problems in a fast and accurate way. The same code runs on major distributed environment (Hadoop, SGE, MPI) and can solve problems beyond billions of examples.", "_____no_output_____" ] ], [ [ "!pip install lightgbm", "_____no_output_____" ] ], [ [ "<a href=\"https://lightgbm.readthedocs.io/en/latest/\" target=\"_blank\">LightGBM</a> is a gradient boosting framework that uses tree based learning algorithms. It is designed to be distributed and efficient with the following advantages:\n* Faster training speed and higher efficiency.\n* Lower memory usage.\n* Better accuracy.\n* Support of parallel and GPU learning.\n* Capable of handling large-scale data.", "_____no_output_____" ] ], [ [ "!pip install altair", "_____no_output_____" ] ], [ [ "<a href=\"https://pypi.org/project/altair/\" target=\"_blank\">Altair</a> is a declarative statistical visualization library for Python. With Altair, you can spend more time understanding your data and its meaning. Altair's API is simple, friendly and consistent and built on top of the powerful Vega-Lite JSON specification. This elegant simplicity produces beautiful and effective visualizations with a minimal amount of code. Altair is developed by Jake Vanderplas and Brian Granger in close collaboration with the UW Interactive Data Lab.", "_____no_output_____" ] ], [ [ "!pip install folium", "_____no_output_____" ] ], [ [ "<a href=\"https://pypi.org/project/folium/\" target=\"_blank\">folium</a> builds on the data wrangling strengths of the Python ecosystem and the mapping strengths of the Leaflet.js library. Manipulate your data in Python, then visualize it in a Leaflet map via folium.", "_____no_output_____" ] ], [ [ "!pip install fbprophet", "_____no_output_____" ] ], [ [ "<a href=\"https://pypi.org/project/fbprophet/\" target=\"_blank\">Prophet</a> is a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. It works best with time series that have strong seasonal effects and several seasons of historical data. Prophet is robust to missing data and shifts in the trend, and typically handles outliers well.", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Umami notebooks\n\nWelcome to the umami notebooks. This page provides links to notebooks that provide an introduction to umami and its use. We recommend that you look at them in the following order.\n\nFirst, look at two notebooks designed to introduce you to the core classes and methods of umami.\n\n * [Part 1: Introduction to umami and the `Metric` class](IntroductionToMetric.ipynb)\n * [Part 2: Introduction to the `Residual` class](IntroductionToResidual.ipynb)\n * [Part 3: Other IO options (using umami without Landlab or terrainbento)](OtherIO_options.ipynb)\n \nThen, look at two notebooks with example applications.\n \n * [Part 4: Example Application](ExampleApplication.ipynb)\n * [Part 5: Application using the Discretized Misfit calculation](DiscretizedMisfit.ipynb)\n \nIf you have comments or questions about the notebooks, the best place to get help is through [GitHub Issues](https://github.com/TerrainBento/umami/issues).\n\n## Some background\n\n### What is umami?\n\nUmami is a package for calculating objective functions or objective function components for Earth surface dynamics modeling. It was designed to work well with [terrainbento](https://github.com/TerrainBento/terrainbento) and other models built with the [Landlab Toolkit](https://github.com/landlab/landlab). However, it is not necessary to do modeling with either of these packages to use umami (this is described further in [Part 3 of the notebook series](OtherIO_options.ipynb).\n\nUmami offers two primary classes:\n* a[`Residual`](https://umami.readthedocs.io/en/latest/umami.residual.html#Residual),\nwhich represents the difference between model and data, and \n* a [`Metric`](https://umami.readthedocs.io/en/latest/umami.metric.html),\nwhich is a calculated value on either model or data. \n\nThe set of currently supported calculations can be found in the [`umami.calculations`](https://umami.readthedocs.io/en/latest/umami.calculations.html) submodule.\n\n### What does it do well?\n\nUmami was designed to provide an input-file based interface for calculating single-value landscape metrics for use in model analysis. This supports reproducible analysis and systematic variation in metric construction. When used with `terrainbento`, one input file can describe the model run, and one input file can describe the model assessment or model-data comparison. This streamlines model analysis applications. Umami also provides multiple output formats (YAML and Dakota), the latter of which is designed to interface with Sandia National Laboratory's [Dakota package](https://dakota.sandia.gov).", "_____no_output_____" ] ] ]

[ "markdown" ]

[ [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "### Mit kellene tudni?\n#### 1. Megfogalmazni egy programozási problémát <!-- .element: class=\"fragment\" -->\n", "_____no_output_____" ], [ "#### 1. Számításelmélet értelmét elmagyarázni <!-- .element: class=\"fragment\" -->\n", "_____no_output_____" ], [ "#### 1. Lebontani egy komplex problémát egyszerűbbekre <!-- .element: class=\"fragment\" -->\n", "_____no_output_____" ], [ "#### 1. Megérteni egy leírt programot <!-- .element: class=\"fragment\" -->\n", "_____no_output_____" ], [ "#### 1. Megírni egy programot <!-- .element: class=\"fragment\" -->\n", "_____no_output_____" ], [ "#### 1. Találni érdeklődést és folytatást <!-- .element: class=\"fragment\" -->\n", "_____no_output_____" ], [ "<center><img src=\"https://pics.me.me/doctors-googling-stuff-online-does-not-make-you-a-doctor-61282088.png\"></center> -->", "_____no_output_____" ], [ "### Deklaratív tudás\nA \"mit?\" kérdésre adott válasz egy probléma kapcsán", "_____no_output_____" ], [ "### Imperatív tudás\nA \"hogyan\" kérdésre adott válasz egy probléma kapcsán", "_____no_output_____" ], [ "### Deklaratív példák\n- tervrajz\n", "_____no_output_____" ], [ "- térképjelzés\n", "_____no_output_____" ], [ "- anatómiai ábra\n", "_____no_output_____" ], [ "### Imperatív példák\n- recept\n", "_____no_output_____" ], [ "- IKEA útmutató\n", "_____no_output_____" ], [ "- útvonalterv\n", "_____no_output_____" ], [ "### Mit adunk át a gépnek?\nEgy probléma megoldására szeretnénk megtanítani/megkérni:\n", "_____no_output_____" ], [ "tetszőleges $x$-re, számolja ki $ \\sqrt{x} $-et úgy, hogy csak osztani, szorozni, meg összeadni tud", "_____no_output_____" ], [ "### Tudás Gyökvonásról\n- deklaratív tudás:\n", "_____no_output_____" ], [ "- $ \\sqrt{x} = y \\Leftrightarrow y^2=x $\n", "_____no_output_____" ], [ "- $ x < y^2 \\Leftrightarrow \\sqrt{x} < y $\n", "_____no_output_____" ], [ "- $ ( (y_1 < x \\land y_2 < x) \\lor (y_1 > x \\land y_2 > x) ) \\Rightarrow $\n", "_____no_output_____" ], [ "- $ \\Rightarrow ( |x - y_1| < |x - y_2| \\Leftrightarrow |x^2 - y_1^2| < |x^2 - y_2^2| ) $\n", "_____no_output_____" ], [ "- imperatív tudás:\n", "_____no_output_____" ], [ "- tippelünk egy kezdeti $ G $ -t, és ameddig nem elég jó, $ G' = \\frac{G+\\frac{x}{G}}{2} $ -vel közelítünk, így tetszőlegesen közel kerülünk az eredményhez\n", "_____no_output_____" ], [ "(minden nemnegatív számok esetén)", "_____no_output_____" ], [ "### Ember\nokos, kreatív, tud következtetni, memóriája ködös,\n", "_____no_output_____" ], [ "szeretné minél kevesebbet nyomkodni a számológépet\n", "_____no_output_____" ], [ "- felhasznál csomó meglévő deklaratív tudást\n", "_____no_output_____" ], [ "- összeállítja egy komplex tippelési folyamattá\n", "_____no_output_____" ], [ "- tippek eredményéből továbbfejleszti a folyamatot\n", "_____no_output_____" ], [ "- felhasznál olyan komplex fogalmakat mint nagyságrend, vagy előjelváltás", "_____no_output_____" ], [ "### Gép\nbuta, nincs önálló gondolata, memóriája tökéletes\n", "_____no_output_____" ], [ "másodpercenként töbmilliárd műveletet el tud végezni,\n", "_____no_output_____" ], [ "- tippel egy nagyon rosszat, pl G = 1\n", "_____no_output_____" ], [ "- valaki megtanította neki hogy $G' = \\frac{G+\\frac{x}{G}}{2}$ közelít, szóval elkezdi\n", "_____no_output_____" ], [ "- Megoldandó probléma: $ \\sqrt{1366561} = ? $\n", "_____no_output_____" ], [ "### Specifikáció\nEgyfajta deklaratív leírása egy programnak\n", "_____no_output_____" ], [ "Miből, mit?\n", "_____no_output_____" ], [ "- állapottér - bemenő/kimenő adatok\n", "_____no_output_____" ], [ "- előfeltétel - amit tudunk a bemenetről\n", "_____no_output_____" ], [ "- utófeltétel - az eredmény kiszámítási szabálya (amit tudunk az eredményről)\n", "_____no_output_____" ], [ "legyen **teljes** és **részletes**", "_____no_output_____" ], [ "### Specifikáció I. példa\nValós számok halmazán megoldható másodfokú egyneletet oldjon meg a program, ami $ax^2+bx+c=0$ formában van megadva\n", "_____no_output_____" ], [ "- Á: $(a,b,c,x_1,x_2 \\in \\mathbb{R})$ <!-- .element: class=\"fragment\" data-fragment-index=\"1\" -->\n", "_____no_output_____" ], [ "- Ef: $(a \\neq 0 \\land b^2-4ac \\geq 0)$ <!-- .element: class=\"fragment\" data-fragment-index=\"2\" -->\n", "_____no_output_____" ], [ "- Uf: $ (x_1 = \\frac{-b + \\sqrt{b^2-4ac}}{2a} \\land x_2 = \\frac{-b - \\sqrt{b^2-4ac}}{2a})$ <!-- .element: class=\"fragment\" data-fragment-index=\"3\" -->", "_____no_output_____" ], [ "### Specifikáció példa 2\nValós számok halmazán megoldható másodfokú egyneletet oldjon meg a program, ami $ax^2+bx+c=0$ formában van megadva\n", "_____no_output_____" ], [ "- Á: $(a,b,c,x_1,x_2 \\in \\mathbb{R})$\n", "_____no_output_____" ], [ "- Ef: $(a \\neq 0 \\land b^2-4ac \\geq 0)$\n", "_____no_output_____" ], [ "- Uf: $ (\\forall i (ax_i^2+bx_i+c=0) \\land ( x_1 = x_2 \\iff b^2-4ac = 0 ))$", "_____no_output_____" ], [ "### Specifikáció teljessége\nAmit az előfeltétel megenged az állapottéren belül, azt az utófeltételnek kezelnie kell\n", "_____no_output_____" ], [ "Pl:\n", "_____no_output_____" ], [ "- Válassza ki két szám közül azt amelyiknek több prím osztója van\n", "_____no_output_____" ], [ "- több különböző vagy több összesen?\n", "_____no_output_____" ], [ "- mi van ha ugyanannyi?\n", "_____no_output_____" ], [ "- Válassza ki egy számsorból azt az 5 számot ami legjobban eltér az átlagtól\n", "_____no_output_____" ], [ "- mi van ha nincs 5 szám a számsorban?\n", "_____no_output_____" ], [ "- mi van ha minden szám ugyanaz?\n", "_____no_output_____" ], [ "- Sakktáblán szereplő állásról döntse el hogy matt-e.\n", "_____no_output_____" ], [ "- mi van ha mindkét fél sakkban van?\n", "_____no_output_____" ], [ "- lehet 11 királynő valakinél?", "_____no_output_____" ], [ "### Specifikáció részletessége\nA program főzzön rántottát:\n", "_____no_output_____" ], [ "- Á: (konyha)\n", "_____no_output_____" ], [ "- Ef: ( )\n", "_____no_output_____" ], [ "- Uf: (ha a konyha alkalmas ennek elkészítésére, a konyhában legyen egy finom rántotta)\n", "_____no_output_____" ], [ "vagy:\n", "_____no_output_____" ], [ "- Á: (serpenyő, főzőlap, 3 tojás, olaj, só, fakanál, tányér)\n", "_____no_output_____" ], [ "- Ef: (a serpenyő, tányér és fakanál tiszta, a tojás nem romlott, a főzőlap képes a serpenyőt 200 Celsius fokra felmelegíteni)\n", "_____no_output_____" ], [ "- Uf: (a tojások összekeverve kerültek a 200 fokos serpenyőbe az olaj után, több mint 4 de kevesebb mint 10 percet töltöttek ott, ebből 20 másodpercnél többet nem álltak kavarás nélkül, végül megsózva kikerültek a tányérra)", "_____no_output_____" ], [ "### Specifikáció II. példa\nKét természetes szám legnagyobb közös osztójának megtalálása\n", "_____no_output_____" ], [ "- Á: $(a,b,x \\in \\mathbb{N})$ <!-- .element: class=\"fragment\" data-fragment-index=\"1\" -->\n", "_____no_output_____" ], [ "- Ef: $(a \\neq 0 \\land b \\neq 0)$ <!-- .element: class=\"fragment\" data-fragment-index=\"2\" -->\n", "_____no_output_____" ], [ "- Uf: $ (x|a \\land x|b \\land \\nexists i(i|a \\land i|b \\land i > x))$ <!-- .element: class=\"fragment\" data-fragment-index=\"3\" -->", "_____no_output_____" ], [ "<a style='text-decoration:none;line-height:16px;display:flex;color:#5B5B62;padding:10px;justify-content:end;' href='https://deepnote.com?utm_source=created-in-deepnote-cell&projectId=978e47b7-a961-4dca-a945-499e8b781a34' target=\"_blank\">\n<img alt='Created in deepnote.com' style='display:inline;max-height:16px;margin:0px;margin-right:7.5px;' src='data:image/svg+xml;base64,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' > </img>\nCreated in <span style='font-weight:600;margin-left:4px;'>Deepnote</span></a>", "_____no_output_____" ] ] ]

[ "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Análise do eleitorado brasileiro\n\nFonte -> http://www.tse.jus.br/eleicoes/estatisticas/estatisticas-eleitorais", "_____no_output_____" ] ], [ [ "# importando as bibliotecas\nimport pandas as pd", "_____no_output_____" ], [ "# Carregando o arquivo csv\ndf = pd.read_csv('eleitorado_municipio_2020.csv', encoding='latin1', sep=';')\n\ndf.head().T", "_____no_output_____" ], [ "# Tamanho do arquivo\ndf.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 5568 entries, 0 to 5567\nData columns (total 43 columns):\n # Column Non-Null Count Dtype \n--- ------ -------------- ----- \n 0 NR_ANO_ELEICAO 5568 non-null int64 \n 1 CD_PAIS 5568 non-null int64 \n 2 NM_PAIS 5568 non-null object\n 3 SG_REGIAO 5568 non-null object\n 4 NM_REGIAO 5568 non-null object\n 5 SG_UF 5568 non-null object\n 6 NM_UF 5568 non-null object\n 7 CD_MUNICIPIO 5568 non-null int64 \n 8 NM_MUNICIPIO 5568 non-null object\n 9 QTD_ELEITORES 5568 non-null int64 \n 10 QTD_ELEITORES_FEMININO 5568 non-null int64 \n 11 QTD_ELEITORES_MASCULINO 5568 non-null int64 \n 12 QTD_ELEITORES_NAOINFORMADO 5568 non-null int64 \n 13 QTD_ELEITORES_MENOR16 5568 non-null int64 \n 14 QTD_ELEITORES_16 5568 non-null int64 \n 15 QTD_ELEITORES_17 5568 non-null int64 \n 16 QTD_ELEITORES_18 5568 non-null int64 \n 17 QTD_ELEITORES_19 5568 non-null int64 \n 18 QTD_ELEITORES_20 5568 non-null int64 \n 19 QTD_ELEITORES_21A24 5568 non-null int64 \n 20 QTD_ELEITORES_25A29 5568 non-null int64 \n 21 QTD_ELEITORES_30A34 5568 non-null int64 \n 22 QTD_ELEITORES_35A39 5568 non-null int64 \n 23 QTD_ELEITORES_40A44 5568 non-null int64 \n 24 QTD_ELEITORES_45A49 5568 non-null int64 \n 25 QTD_ELEITORES_50A54 5568 non-null int64 \n 26 QTD_ELEITORES_55A59 5568 non-null int64 \n 27 QTD_ELEITORES_60A64 5568 non-null int64 \n 28 QTD_ELEITORES_65A69 5568 non-null int64 \n 29 QTD_ELEITORES_70A74 5568 non-null int64 \n 30 QTD_ELEITORES_75A79 5568 non-null int64 \n 31 QTD_ELEITORES_80A84 5568 non-null int64 \n 32 QTD_ELEITORES_85A89 5568 non-null int64 \n 33 QTD_ELEITORES_90A94 5568 non-null int64 \n 34 QTD_ELEITORES_95A99 5568 non-null int64 \n 35 QTD_ELEITORES_MAIORIGUAL100 5568 non-null int64 \n 36 QTD_ELEITORES_IDADEINVALIDO 5568 non-null int64 \n 37 QTD_ELEITORES_IDADENAOSEAPLICA 5568 non-null int64 \n 38 QTD_ELEITORES_IDADENAOINFORMADA 5568 non-null int64 \n 39 QTD_ELEITORES_COMBIOMETRIA 5568 non-null int64 \n 40 QTD_ELEITORES_SEMBIOMETRIA 5568 non-null int64 \n 41 QTD_ELEITORES_DEFICIENTE 5568 non-null int64 \n 42 QUANTITATIVO_NOMESOCIAL 5568 non-null int64 \ndtypes: int64(37), object(6)\nmemory usage: 1.8+ MB\n" ], [ "# As cinco cidades com o maior número de eleitores deficientes.\ndf.nlargest(5,'QTD_ELEITORES_DEFICIENTE')", "_____no_output_____" ], [ "# As cinco cidades com o menor número de eleitores.\ndf.nsmallest(5,'QTD_ELEITORES')", "_____no_output_____" ], [ "# Quantidade de eleitores por gênero\nprint('Eleitoras: ', df['QTD_ELEITORES_FEMININO'].sum())\nprint('Eleitores: ', df['QTD_ELEITORES_MASCULINO'].sum())\nprint('Não informado: ', df['QTD_ELEITORES_NAOINFORMADO'].sum())", "Eleitoras: 77649569\nEleitores: 70228457\nNão informado: 40457\n" ], [ "# Criar variáveis para calcular o percentual\ntot_eleitores = df['QTD_ELEITORES'].sum()\ntot_masc = df['QTD_ELEITORES_MASCULINO'].sum()\ntot_fem = df['QTD_ELEITORES_FEMININO'].sum()\ntot_ninf = df['QTD_ELEITORES_NAOINFORMADO'].sum()", "_____no_output_____" ], [ "# Mostrar valores percentuais\nprint('Eleitoras: ', ( ( tot_fem / tot_eleitores ) * 100) .round(2), '%' ) \nprint('Eleitores: ', ( ( tot_masc / tot_eleitores ) * 100) .round(2), '%' ) \nprint('Não informado: ', ( ( tot_ninf / tot_eleitores ) * 100) .round(2), '%' ) ", "Eleitoras: 52.49 %\nEleitores: 47.48 %\nNão informado: 0.03 %\n" ], [ "# Quantos municípios com mais homens do que mulheres\ndf[df['QTD_ELEITORES_MASCULINO'] > df['QTD_ELEITORES_FEMININO']].count()", "_____no_output_____" ], [ "# Vamos criar uma coluna nova para indicar a relação fem/masc\ndf['RELACAO_FM'] = df['QTD_ELEITORES_FEMININO'] / df['QTD_ELEITORES_MASCULINO']", "_____no_output_____" ], [ "# Quais os municípios com maior relação fem/masc?\ndf.nlargest(5,'RELACAO_FM').T", "_____no_output_____" ], [ "# Quais os municípios com menor relação fem/masc?\ndf.nsmallest(5,'RELACAO_FM')", "_____no_output_____" ], [ "# Vamos criar um DataFrame só com municípios de São Paulo\ndf_sp = df[ df['SG_UF'] == 'SP' ].copy()\ndf_sp.head()", "_____no_output_____" ], [ "# Quais os municípios com maior relação fem/masc de São Paulo?\ndf_sp.nlargest(5,'RELACAO_FM')", "_____no_output_____" ], [ "# Quais os municípios com menor relação fem/masc de São Paulo?\ndf_sp.nsmallest(5,'RELACAO_FM')", "_____no_output_____" ], [ "# Plotar um gráfico de distribuição fem/masc\ndf['RELACAO_FM'].plot.hist(bins=100)", "_____no_output_____" ], [ "# carregar biblioteca gráfica\nimport seaborn as sns\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "# Plotar um gráfico de distribuição fem/masc\nsns.distplot( df['RELACAO_FM'] , bins=100 , color='red', kde=False )\n\n# Embelezando o gráfico\nplt.title('Relação Eleitoras/Eleitores', fontsize=18)\nplt.xlabel('Eleitoras/Eleitores', fontsize=14)\nplt.ylabel('Frequência', fontsize=14)\nplt.axvline(1.0, color='black', linestyle='--')", "_____no_output_____" ], [ "# Plotar um gráfico de distribuição fem/masc, mas mostrando os pontos (municípios)\nsns.swarmplot( data=df , x='NM_REGIAO', y='RELACAO_FM' )\nplt.axhline(1.0, color='black', linestyle='--')", "_____no_output_____" ], [ "# Vamos plotar o gráfico de total de eleitores por faixa etária usando o gráfico de barras horizontal\n\n# primeiro vamos listar as colunas de nosso interesse\nlista = ['QTD_ELEITORES_16', 'QTD_ELEITORES_17', 'QTD_ELEITORES_18', 'QTD_ELEITORES_19', 'QTD_ELEITORES_20']\n\ntot_idade = df[lista].sum()", "_____no_output_____" ], [ "tot_idade", "_____no_output_____" ], [ "# Mostrando o gráfico de barras\n\ntot_idade.plot.barh()", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Gun Deaths in the US, 2012-2014\n\nThis is an analysis of gun deaths in the US between the years 2012 and 2014. The data comes from FiveThirtyEight, specifcally [here](https://github.com/fivethirtyeight/guns-data).", "_____no_output_____" ] ], [ [ "import csv\n\nf = open(\"guns.csv\", \"r\")\ndata = list(csv.reader(f))\nprint(data[:5])", "[['', 'year', 'month', 'intent', 'police', 'sex', 'age', 'race', 'hispanic', 'place', 'education'], ['1', '2012', '01', 'Suicide', '0', 'M', '34', 'Asian/Pacific Islander', '100', 'Home', '4'], ['2', '2012', '01', 'Suicide', '0', 'F', '21', 'White', '100', 'Street', '3'], ['3', '2012', '01', 'Suicide', '0', 'M', '60', 'White', '100', 'Other specified', '4'], ['4', '2012', '02', 'Suicide', '0', 'M', '64', 'White', '100', 'Home', '4']]\n" ], [ "headers = data[0]\ndata = data[1:]\nprint(headers)\nprint(data[:5])", "['', 'year', 'month', 'intent', 'police', 'sex', 'age', 'race', 'hispanic', 'place', 'education']\n[['1', '2012', '01', 'Suicide', '0', 'M', '34', 'Asian/Pacific Islander', '100', 'Home', '4'], ['2', '2012', '01', 'Suicide', '0', 'F', '21', 'White', '100', 'Street', '3'], ['3', '2012', '01', 'Suicide', '0', 'M', '60', 'White', '100', 'Other specified', '4'], ['4', '2012', '02', 'Suicide', '0', 'M', '64', 'White', '100', 'Home', '4'], ['5', '2012', '02', 'Suicide', '0', 'M', '31', 'White', '100', 'Other specified', '2']]\n" ], [ "import datetime\n\ndates = [datetime.datetime(year=int(d[1]), month=int(d[2]), day=1) for d in data]\nprint(dates[:5])\n\ndate_counts = {}\nfor d in dates:\n if d in date_counts:\n date_counts[d] += 1\n else:\n date_counts[d] = 1\nprint(date_counts)", "[datetime.datetime(2012, 1, 1, 0, 0), datetime.datetime(2012, 1, 1, 0, 0), datetime.datetime(2012, 1, 1, 0, 0), datetime.datetime(2012, 2, 1, 0, 0), datetime.datetime(2012, 2, 1, 0, 0)]\n{datetime.datetime(2012, 3, 1, 0, 0): 2743, datetime.datetime(2014, 8, 1, 0, 0): 2970, datetime.datetime(2014, 2, 1, 0, 0): 2361, datetime.datetime(2014, 7, 1, 0, 0): 2884, datetime.datetime(2014, 4, 1, 0, 0): 2862, datetime.datetime(2014, 6, 1, 0, 0): 2931, datetime.datetime(2012, 6, 1, 0, 0): 2826, datetime.datetime(2012, 11, 1, 0, 0): 2729, datetime.datetime(2014, 9, 1, 0, 0): 2914, datetime.datetime(2014, 3, 1, 0, 0): 2684, datetime.datetime(2014, 1, 1, 0, 0): 2651, datetime.datetime(2013, 10, 1, 0, 0): 2808, datetime.datetime(2014, 5, 1, 0, 0): 2864, datetime.datetime(2012, 1, 1, 0, 0): 2758, datetime.datetime(2012, 10, 1, 0, 0): 2733, datetime.datetime(2014, 10, 1, 0, 0): 2865, datetime.datetime(2013, 5, 1, 0, 0): 2806, datetime.datetime(2013, 3, 1, 0, 0): 2862, datetime.datetime(2014, 11, 1, 0, 0): 2756, datetime.datetime(2013, 1, 1, 0, 0): 2864, datetime.datetime(2012, 12, 1, 0, 0): 2791, datetime.datetime(2012, 4, 1, 0, 0): 2795, datetime.datetime(2012, 8, 1, 0, 0): 2954, datetime.datetime(2013, 7, 1, 0, 0): 3079, datetime.datetime(2012, 2, 1, 0, 0): 2357, datetime.datetime(2013, 6, 1, 0, 0): 2920, datetime.datetime(2012, 7, 1, 0, 0): 3026, datetime.datetime(2013, 11, 1, 0, 0): 2758, datetime.datetime(2012, 9, 1, 0, 0): 2852, datetime.datetime(2014, 12, 1, 0, 0): 2857, datetime.datetime(2012, 5, 1, 0, 0): 2999, datetime.datetime(2013, 2, 1, 0, 0): 2375, datetime.datetime(2013, 12, 1, 0, 0): 2765, datetime.datetime(2013, 9, 1, 0, 0): 2742, datetime.datetime(2013, 8, 1, 0, 0): 2859, datetime.datetime(2013, 4, 1, 0, 0): 2798}\n" ], [ "sex_counts = {}\nfor d in data:\n if d[5] in sex_counts:\n sex_counts[d[5]] += 1\n else:\n sex_counts[d[5]] = 1\n\nrace_counts = {}\nfor d in data:\n if d[7] in race_counts:\n race_counts[d[7]] += 1\n else:\n race_counts[d[7]] = 1\n\nprint(sex_counts)\nprint(race_counts)", "{'F': 14449, 'M': 86349}\n{'Native American/Native Alaskan': 917, 'Hispanic': 9022, 'White': 66237, 'Asian/Pacific Islander': 1326, 'Black': 23296}\n" ] ], [ [ "## Patterns so far\n* Most gun deaths are for race = white and sex = M, but is there a further correlation here?\n* It may help to explore this further to find more patterns like age, education status, and if police were involved", "_____no_output_____" ] ], [ [ "f2 = open(\"census.csv\", \"r\")\ncensus = list(csv.reader(f2))\ncensus", "_____no_output_____" ], [ "mapping = {\"Asian/Pacific Islander\": 15834141, \n \"Black\": 40250635, \n \"Native American/Native Alaskan\": 3739506, \n \"Hispanic\": 44618105, \n \"White\": 197318956}\nrace_per_hundredk = {}\nfor r, v in race_counts.items():\n race_per_hundredk[r] = (v / mapping[r]) * 100000\nprint(race_per_hundredk)", "{'Native American/Native Alaskan': 24.521955573811088, 'Hispanic': 20.220491210910907, 'White': 33.56849303419181, 'Asian/Pacific Islander': 8.374309664161762, 'Black': 57.8773477735196}\n" ], [ "intent = [d[3] for d in data]\nraces = [d[7] for d in data]\nhomicide_race_counts = {}\nfor i, race in enumerate(races):\n if intent[i] == \"Suicide\":\n if race in homicide_race_counts:\n homicide_race_counts[race] += 1\n else:\n homicide_race_counts[race] = 1\nhomicide_race_per_hundredk = {}\nfor r, v in homicide_race_counts.items():\n homicide_race_per_hundredk[r] = (v / mapping[r]) * 100000\nprint(homicide_race_per_hundredk)", "{'Native American/Native Alaskan': 14.841532544673013, 'Hispanic': 7.106980451097149, 'White': 28.06217969245692, 'Asian/Pacific Islander': 4.705023152187416, 'Black': 8.278130270491385}\n" ] ], [ [ "# More findings\n* If Homicide is the intent, the black race is significantly more likely to be affected by gun death (over 10 times that of whites and 4 times that of Hispanic)\n* If Suicide is the intent, the white race is significantly more likely to be affected by gun death (2 times that of Native American and over 3 times that of Black or Hispanic)\n\n# Next steps\n* Investigate further if/where age comes into play as well as education", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Módulo 2: Scraping con Selenium\n## LATAM Airlines\n<a href=\"https://www.latam.com/es_ar/\"><img src=\"https://i.pinimg.com/originals/dd/52/74/dd5274702d1382d696caeb6e0f6980c5.png\" width=\"420\"></img></a>\n<br>\n\nVamos a scrapear el sitio de Latam para averiguar datos de vuelos en funcion el origen y destino, fecha y cabina. La información que esperamos obtener de cada vuelo es:\n- Precio(s) disponibles\n- Horas de salida y llegada (duración)\n- Información de las escalas\n\n¡Empecemos!", "_____no_output_____" ] ], [ [ "url = 'https://www.latam.com/es_ar/apps/personas/booking?fecha1_dia=20&fecha1_anomes=2019-12&auAvailability=1&ida_vuelta=ida&vuelos_origen=Buenos%20Aires&from_city1=BUE&vuelos_destino=Madrid&to_city1=MAD&flex=1&vuelos_fecha_salida_ddmmaaaa=20/12/2019&cabina=Y&nadults=1&nchildren=0&ninfants=0&cod_promo='", "_____no_output_____" ], [ "from selenium import webdriver", "_____no_output_____" ], [ "options = webdriver.ChromeOptions()\noptions.add_argument('--incognito')\ndriver = webdriver.Chrome(executable_path='../../chromedriver', options=options)\ndriver.get(url)", "_____no_output_____" ], [ "#Usaremos el Xpath para obtener la lista de vuelos\nvuelos = driver.find_elements_by_xpath('//li[@class=\"flight\"]')", "_____no_output_____" ], [ "vuelo = vuelos[0]", "_____no_output_____" ] ], [ [ "Obtenemos la información de la hora de salida, llegada y duración del vuelo", "_____no_output_____" ] ], [ [ "# Hora de salida\nvuelo.find_element_by_xpath('.//div[@class=\"departure\"]/time').get_attribute('datetime')", "_____no_output_____" ], [ "# Hora de llegada\nvuelo.find_element_by_xpath('.//div[@class=\"arrival\"]/time').get_attribute('datetime')", "_____no_output_____" ], [ "# Duración del vuelo\nvuelo.find_element_by_xpath('.//span[@class=\"duration\"]/time').get_attribute('datetime')", "_____no_output_____" ], [ "boton_escalas = vuelo.find_element_by_xpath('.//div[@class=\"flight-summary-stops-description\"]/button')\nboton_escalas", "_____no_output_____" ], [ "boton_escalas.click()", "_____no_output_____" ], [ "segmentos = vuelo.find_elements_by_xpath('//div[@class=\"segments-graph\"]/div[@class=\"segments-graph-segment\"]')\nsegmentos", "_____no_output_____" ], [ "escalas = len(segmentos) - 1 #0 escalas si es un vuelo directo", "_____no_output_____" ], [ "segmento = segmentos[0]", "_____no_output_____" ], [ "# Origen\nsegmento.find_element_by_xpath('.//div[@class=\"departure\"]/span[@class=\"ground-point-name\"]').text", "_____no_output_____" ], [ "# Hora de salida\nsegmento.find_element_by_xpath('.//div[@class=\"departure\"]/time').get_attribute('datetime')", "_____no_output_____" ], [ "# Destino\nsegmento.find_element_by_xpath('.//div[@class=\"arrival\"]/span[@class=\"ground-point-name\"]').text", "_____no_output_____" ], [ "# Hora de llegada\nsegmento.find_element_by_xpath('.//div[@class=\"arrival\"]/time').get_attribute('datetime')", "_____no_output_____" ], [ "# Duración del vuelo\nsegmento.find_element_by_xpath('.//span[@class=\"duration flight-schedule-duration\"]/time').get_attribute('datetime')", "_____no_output_____" ], [ "# Numero del vuelo\nsegmento.find_element_by_xpath('.//span[@class=\"equipment-airline-number\"]').text", "_____no_output_____" ], [ "# Modelo de avion\nsegmento.find_element_by_xpath('.//span[@class=\"equipment-airline-material\"]').text", "_____no_output_____" ], [ "# Duracion de la escala\nsegmento.find_element_by_xpath('.//div[@class=\"stop connection\"]//p[@class=\"stop-wait-time\"]//time').get_attribute('datetime')", "_____no_output_____" ], [ "vuelo.find_element_by_xpath('//div[@class=\"modal-dialog\"]//button[@class=\"close\"]').click()", "_____no_output_____" ], [ "vuelo.click()", "_____no_output_____" ], [ "tarifas = vuelo.find_elements_by_xpath('.//div[@class=\"fares-table-container\"]//tfoot//td[contains(@class, \"fare-\")]')", "_____no_output_____" ], [ "precios = []\nfor tarifa in tarifas:\n nombre = tarifa.find_element_by_xpath('.//label').get_attribute('for')\n moneda = tarifa.find_element_by_xpath('.//span[@class=\"price\"]/span[@class=\"currency-symbol\"]').text\n valor = tarifa.find_element_by_xpath('.//span[@class=\"price\"]/span[@class=\"value\"]').text \n dict_tarifa={nombre:{'moneda':moneda, 'valor':valor}}\n precios.append(dict_tarifa)\n print(dict_tarifa)", "_____no_output_____" ], [ "def obtener_tiempos(vuelo):\n # Hora de salida\n salida = vuelo.find_element_by_xpath('.//div[@class=\"departure\"]/time').get_attribute('datetime')\n # Hora de llegada\n llegada = vuelo.find_element_by_xpath('.//div[@class=\"arrival\"]/time').get_attribute('datetime')\n # Duracion\n duracion = vuelo.find_element_by_xpath('.//span[@class=\"duration\"]/time').get_attribute('datetime')\n return {'hora_salida': salida, 'hora_llegada': llegada, 'duracion': duracion}", "_____no_output_____" ], [ "def obtener_precios(vuelo):\n tarifas = vuelo.find_elements_by_xpath(\n './/div[@class=\"fares-table-container\"]//tfoot//td[contains(@class, \"fare-\")]')\n precios = []\n for tarifa in tarifas:\n nombre = tarifa.find_element_by_xpath('.//label').get_attribute('for')\n moneda = tarifa.find_element_by_xpath('.//span[@class=\"price\"]/span[@class=\"currency-symbol\"]').text\n valor = tarifa.find_element_by_xpath('.//span[@class=\"price\"]/span[@class=\"value\"]').text \n dict_tarifa={nombre:{'moneda':moneda, 'valor':valor}}\n precios.append(dict_tarifa)\n return precios", "_____no_output_____" ], [ "def obtener_datos_escalas(vuelo):\n segmentos = vuelo.find_elements_by_xpath('//div[@class=\"segments-graph\"]/div[@class=\"segments-graph-segment\"]')\n info_escalas = []\n for segmento in segmentos:\n # Origen\n origen = segmento.find_element_by_xpath(\n './/div[@class=\"departure\"]/span[@class=\"ground-point-name\"]').text\n # Hora de salida\n dep_time = segmento.find_element_by_xpath(\n './/div[@class=\"departure\"]/time').get_attribute('datetime')\n # Destino\n destino = segmento.find_element_by_xpath(\n './/div[@class=\"arrival\"]/span[@class=\"ground-point-name\"]').text\n # Hora de llegada\n arr_time = segmento.find_element_by_xpath(\n './/div[@class=\"arrival\"]/time').get_attribute('datetime')\n # Duración del vuelo\n duracion_vuelo = segmento.find_element_by_xpath(\n './/span[@class=\"duration flight-schedule-duration\"]/time').get_attribute('datetime')\n # Numero del vuelo\n numero_vuelo = segmento.find_element_by_xpath(\n './/span[@class=\"equipment-airline-number\"]').text\n # Modelo de avion\n modelo_avion = segmento.find_element_by_xpath(\n './/span[@class=\"equipment-airline-material\"]').text\n # Duracion de la escala\n if segmento != segmentos[-1]:\n duracion_escala = segmento.find_element_by_xpath(\n './/div[@class=\"stop connection\"]//p[@class=\"stop-wait-time\"]//time').get_attribute('datetime')\n else:\n duracion_escala = ''\n\n # Armo un diccionario para almacenar los datos\n data_dict={'origen': origen, \n 'dep_time': dep_time, \n 'destino': destino,\n 'arr_time': arr_time,\n 'duracion_vuelo': duracion_vuelo,\n 'numero_vuelo': numero_vuelo,\n 'modelo_avion': modelo_avion,\n 'duracion_escala': duracion_escala}\n info_escalas.append(data_dict)\n \n return info_escalas", "_____no_output_____" ] ], [ [ "## Clase 15\nYa tenemos el scraper casi listo. Unifiquemos las 3 funciones de la clase anterior en una sola", "_____no_output_____" ] ], [ [ "def obtener_info(driver):\n vuelos = driver.find_elements_by_xpath('//li[@class=\"flight\"]')\n print(f'Se encontraron {len(vuelos)} vuelos.')\n print('Iniciando scraping...')\n info = []\n for vuelo in vuelos:\n # Obtenemos los tiempos generales del vuelo\n tiempos = obtener_tiempos(vuelo)\n # Clickeamos el botón de escalas para ver los detalles\n vuelo.find_element_by_xpath('.//div[@class=\"flight-summary-stops-description\"]/button').click()\n escalas = obtener_datos_escalas(vuelo)\n # Cerramos el pop-up con los detalles\n vuelo.find_element_by_xpath('//div[@class=\"modal-dialog\"]//button[@class=\"close\"]').click()\n # Clickeamos el vuelo para ver los precios\n vuelo.click()\n precios = obtener_precios(vuelo)\n # Cerramos los precios del vuelo\n vuelo.click()\n info.append({'precios':precios, 'tiempos':tiempos , 'escalas': escalas})\n return info", "_____no_output_____" ] ], [ [ "Ahora podemos cargar la página con el driver y pasárselo a esta función", "_____no_output_____" ] ], [ [ "driver = webdriver.Chrome(executable_path='../../chromedriver', options=options)\ndriver.get(url)\nobtener_info(driver)", "_____no_output_____" ] ], [ [ "Se encontraron 0 vuelos porque la página no terminó de cargar", "_____no_output_____" ], [ "Lo más simple que podemos hacer es agregar una demora fija lo suficientemente grande para asegurarnos que la página terminó de cargar.", "_____no_output_____" ] ], [ [ "import time", "_____no_output_____" ], [ "options = webdriver.ChromeOptions()\noptions.add_argument('--incognito')\ndriver = webdriver.Chrome(executable_path='../../chromedriver', options=options)\ndriver.get(url)", "_____no_output_____" ], [ "time.sleep(10)", "_____no_output_____" ], [ "vuelos = driver.find_elements_by_xpath('//li[@class=\"flight\"]')\nvuelos", "_____no_output_____" ], [ "driver.close()", "_____no_output_____" ] ], [ [ "Esto funciona pero no es muy eficiente. Lo mejor sería esperar a que la página termine de cargar y luego recuperar los elementos.", "_____no_output_____" ] ], [ [ "from selenium.webdriver.support.ui import WebDriverWait\nfrom selenium.webdriver.support import expected_conditions as EC\nfrom selenium.webdriver.common.by import By\nfrom selenium.common.exceptions import TimeoutException", "_____no_output_____" ], [ "options = webdriver.ChromeOptions()\noptions.add_argument('--incognito')\ndriver = webdriver.Chrome(executable_path='../../chromedriver', options=options)\ndriver.get(url)\ndelay = 10\ntry:\n vuelo = WebDriverWait(driver, delay).until(EC.presence_of_element_located((By.XPATH, '//li[@class=\"flight\"]')))\n print(\"La página terminó de cargar\")\n info_vuelos = obtener_info(driver)\nexcept TimeoutException:\n print(\"La página tardó demasiado en cargar\")\ndriver.close()", "_____no_output_____" ], [ "info_vuelos", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ] ]

[ "CC0-1.0" ]

[ "CC0-1.0" ]

[ "CC0-1.0" ]

[ [ [ "# Importing the required libraries\nimport math\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "# Importing the dataset\ndf=pd.read_csv('Ads_CTR_Optimisation.csv')", "_____no_output_____" ], [ "# Implementing Upper confidence bound model\nN=10000\nd=10\ntotal_ads_selected=[]\ntotal_selections=[0]*d\nreward_sum=[0]*d\ntotal_reward=0\nfor n in range(0,N):\n ad=0\n max_upper_bound=0\n for i in range(0,d):\n if (total_selections[i]>0):\n avg_reward=reward_sum[i]/total_selections[i]\n deltai=math.sqrt(3/2*math.log(n+1)/total_selections[i])\n upper_bound=avg_reward+deltai\n else:\n upper_bound=1e400\n if (upper_bound>max_upper_bound):\n max_upper_bound=upper_bound\n ad=i\n total_ads_selected.append(ad)\n total_selections[ad]=total_selections[ad]+1\n reward=df.values[n,ad]\n reward_sum[ad]=reward_sum[ad]+reward\n total_reward=total_reward+reward", "_____no_output_____" ], [ "# Visualising the results\n\nplt.hist(total_ads_selected)\nplt.xlabel('Advertisements')\nplt.ylabel('Total number of times each ad was clicked')\nplt.show()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## RIHAD VARIAWA, Data Scientist - Who has fun LEARNING, EXPLORING & GROWING\n<h1>2D <code>Numpy</code> in Python</h1>", "_____no_output_____" ], [ "<p><strong>Welcome!</strong> This notebook will teach you about using <code>Numpy</code> in the Python Programming Language. By the end of this lab, you'll know what <code>Numpy</code> is and the <code>Numpy</code> operations.</p>", "_____no_output_____" ], [ "<h2>Table of Contents</h2>\n<div class=\"alert alert-block alert-info\" style=\"margin-top: 20px\">\n <ul>\n <li><a href=\"create\">Create a 2D Numpy Array</a></li>\n <li><a href=\"access\">Accessing different elements of a Numpy Array</a></li>\n <li><a href=\"op\">Basic Operations</a></li>\n </ul>\n <p>\n Estimated time needed: <strong>20 min</strong>\n </p>\n</div>\n\n<hr>", "_____no_output_____" ], [ "<h2 id=\"create\">Create a 2D Numpy Array</h2>", "_____no_output_____" ] ], [ [ "# Import the libraries\n\nimport numpy as np \nimport matplotlib.pyplot as plt", "_____no_output_____" ] ], [ [ "Consider the list <code>a</code>, the list contains three nested lists **each of equal size**. ", "_____no_output_____" ] ], [ [ "# Create a list\n\na = [[11, 12, 13], [21, 22, 23], [31, 32, 33]]\na", "_____no_output_____" ] ], [ [ "We can cast the list to a Numpy Array as follow", "_____no_output_____" ] ], [ [ "# Convert list to Numpy Array\n# Every element is the same type\n\nA = np.array(a)\nA", "_____no_output_____" ] ], [ [ "We can use the attribute <code>ndim</code> to obtain the number of axes or dimensions referred to as the rank. ", "_____no_output_____" ] ], [ [ "# Show the numpy array dimensions\n\nA.ndim", "_____no_output_____" ] ], [ [ "Attribute <code>shape</code> returns a tuple corresponding to the size or number of each dimension.", "_____no_output_____" ] ], [ [ "# Show the numpy array shape\n\nA.shape", "_____no_output_____" ] ], [ [ "The total number of elements in the array is given by the attribute <code>size</code>.", "_____no_output_____" ] ], [ [ "# Show the numpy array size\n\nA.size", "_____no_output_____" ] ], [ [ "<hr>", "_____no_output_____" ], [ "<h2 id=\"access\">Accessing different elements of a Numpy Array</h2>", "_____no_output_____" ], [ "We can use rectangular brackets to access the different elements of the array. The correspondence between the rectangular brackets and the list and the rectangular representation is shown in the following figure for a 3x3 array: ", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoEg.png\" width=\"500\" />", "_____no_output_____" ], [ "We can access the 2nd-row 3rd column as shown in the following figure:", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoFT.png\" width=\"400\" />", "_____no_output_____" ], [ " We simply use the square brackets and the indices corresponding to the element we would like:", "_____no_output_____" ] ], [ [ "# Access the element on the second row and third column\n\nA[1, 2]", "_____no_output_____" ] ], [ [ " We can also use the following notation to obtain the elements: ", "_____no_output_____" ] ], [ [ "# Access the element on the second row and third column\n\nA[1][2]", "_____no_output_____" ] ], [ [ " Consider the elements shown in the following figure ", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoFF.png\" width=\"400\" />", "_____no_output_____" ], [ "We can access the element as follows ", "_____no_output_____" ] ], [ [ "# Access the element on the first row and first column\n\nA[0][0]", "_____no_output_____" ] ], [ [ "We can also use slicing in numpy arrays. Consider the following figure. We would like to obtain the first two columns in the first row", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoFSF.png\" width=\"400\" />", "_____no_output_____" ], [ " This can be done with the following syntax ", "_____no_output_____" ] ], [ [ "# Access the element on the first row and first and second columns\n\nA[0][0:2]", "_____no_output_____" ] ], [ [ "Similarly, we can obtain the first two rows of the 3rd column as follows:", "_____no_output_____" ] ], [ [ "# Access the element on the first and second rows and third column\n\nA[0:2, 2]", "_____no_output_____" ] ], [ [ "Corresponding to the following figure: ", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoTST.png\" width=\"400\" />", "_____no_output_____" ], [ "<hr>", "_____no_output_____" ], [ "<h2 id=\"op\">Basic Operations</h2>", "_____no_output_____" ], [ "We can also add arrays. The process is identical to matrix addition. Matrix addition of <code>X</code> and <code>Y</code> is shown in the following figure:", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoAdd.png\" width=\"500\" />", "_____no_output_____" ], [ "The numpy array is given by <code>X</code> and <code>Y</code>", "_____no_output_____" ] ], [ [ "# Create a numpy array X\n\nX = np.array([[1, 0], [0, 1]]) \nX", "_____no_output_____" ], [ "# Create a numpy array Y\n\nY = np.array([[2, 1], [1, 2]]) \nY", "_____no_output_____" ] ], [ [ " We can add the numpy arrays as follows.", "_____no_output_____" ] ], [ [ "# Add X and Y\n\nZ = X + Y\nZ", "_____no_output_____" ] ], [ [ "Multiplying a numpy array by a scaler is identical to multiplying a matrix by a scaler. If we multiply the matrix <code>Y</code> by the scaler 2, we simply multiply every element in the matrix by 2 as shown in the figure.", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoDb.png\" width=\"500\" />", "_____no_output_____" ], [ "We can perform the same operation in numpy as follows ", "_____no_output_____" ] ], [ [ "# Create a numpy array Y\n\nY = np.array([[2, 1], [1, 2]]) \nY", "_____no_output_____" ], [ "# Multiply Y with 2\n\nZ = 2 * Y\nZ", "_____no_output_____" ] ], [ [ "Multiplication of two arrays corresponds to an element-wise product or Hadamard product. Consider matrix <code>X</code> and <code>Y</code>. The Hadamard product corresponds to multiplying each of the elements in the same position, i.e. multiplying elements contained in the same color boxes together. The result is a new matrix that is the same size as matrix <code>Y</code> or <code>X</code>, as shown in the following figure.", "_____no_output_____" ], [ "<img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Chapter%205/Images/NumTwoMul.png\" width=\"500\" />", "_____no_output_____" ], [ "We can perform element-wise product of the array <code>X</code> and <code>Y</code> as follows:", "_____no_output_____" ] ], [ [ "# Create a numpy array Y\n\nY = np.array([[2, 1], [1, 2]]) \nY", "_____no_output_____" ], [ "# Create a numpy array X\n\nX = np.array([[1, 0], [0, 1]]) \nX", "_____no_output_____" ], [ "# Multiply X with Y\n\nZ = X * Y\nZ", "_____no_output_____" ] ], [ [ "We can also perform matrix multiplication with the numpy arrays <code>A</code> and <code>B</code> as follows:", "_____no_output_____" ], [ "First, we define matrix <code>A</code> and <code>B</code>:", "_____no_output_____" ] ], [ [ "# Create a matrix A\n\nA = np.array([[0, 1, 1], [1, 0, 1]])\nA", "_____no_output_____" ], [ "# Create a matrix B\n\nB = np.array([[1, 1], [1, 1], [-1, 1]])\nB", "_____no_output_____" ] ], [ [ "We use the numpy function <code>dot</code> to multiply the arrays together.", "_____no_output_____" ] ], [ [ "# Calculate the dot product\n\nZ = np.dot(A,B)\nZ", "_____no_output_____" ], [ "# Calculate the sine of Z\n\nnp.sin(Z)", "_____no_output_____" ] ], [ [ "We use the numpy attribute <code>T</code> to calculate the transposed matrix", "_____no_output_____" ] ], [ [ "# Create a matrix C\n\nC = np.array([[1,1],[2,2],[3,3]])\nC", "_____no_output_____" ], [ "# Get the transposed of C\n\nC.T", "_____no_output_____" ] ], [ [ "<hr>\n<h2>The last exercise!</h2>\n<p>Congratulations, you have completed your first lesson and hands-on lab in Python. However, there is one more thing you need to do. The Data Science community encourages sharing work. The best way to share and showcase your work is to share it on GitHub. By sharing your notebook on GitHub you are not only building your reputation with fellow data scientists, but you can also show it off when applying for a job. Even though this was your first piece of work, it is never too early to start building good habits. So, please read and follow <a href=\"https://cognitiveclass.ai/blog/data-scientists-stand-out-by-sharing-your-notebooks/\" target=\"_blank\">this article</a> to learn how to share your work.\n<hr>", "_____no_output_____" ], [ "<div class=\"alert alert-block alert-info\" style=\"margin-top: 20px\">\n<h2>Get IBM Watson Studio free of charge!</h2>\n <p><a href=\"https://cocl.us/bottemNotebooksPython101Coursera\"><img src=\"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/PY0101EN/Ad/BottomAd.png\" width=\"750\" align=\"center\"></a></p>\n</div>", "_____no_output_____" ], [ "<h3>About the Authors:</h3> \n<p><a href=\"https://www.linkedin.com/in/joseph-s-50398b136/\" target=\"_blank\">Joseph Santarcangelo</a> is a Data Scientist at IBM, and holds a PhD in Electrical Engineering. His research focused on using Machine Learning, Signal Processing, and Computer Vision to determine how videos impact human cognition. Joseph has been working for IBM since he completed his PhD.</p>", "_____no_output_____" ], [ "Other contributors: <a href=\"www.linkedin.com/in/jiahui-mavis-zhou-a4537814a\">Mavis Zhou</a>", "_____no_output_____" ], [ "<hr>", "_____no_output_____" ], [ "<p>Copyright &copy; 2018 IBM Developer Skills Network. This notebook and its source code are released under the terms of the <a href=\"https://cognitiveclass.ai/mit-license/\">MIT License</a>.</p>", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Master Notebook: Bosques aleatorios", "_____no_output_____" ], [ "Como ya vímos en scikit-learn gran parte de codigo es reciclable. Particularmente, leyendo variables y preparando los datos es lo mismo, independientemente del clasificador que usamos.", "_____no_output_____" ], [ "## Leyendo datos", "_____no_output_____" ], [ "Para que no esta tan aburrido (también para mi) esta vez nos vamos a escribir una función para leer los datos que podemos reciclar para otros datos.", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom collections import Counter\n\ndef lea_datos(archivo, i_clase=-1, encabezado=True, delim=\",\"):\n '''Una funcion para leer archivos con datos de clasificación. \n Argumentos:\n archivo - direccion del archivo\n i_clase - indice de columna que contiene las clases. \n default es -1 y significa la ultima fila.\n header - si hay un encabezado\n delim - que separa los datos\n Regresa:\n Un tuple de los datos, clases y cabezazo en caso que hay.'''\n \n todo = np.loadtxt(archivo, dtype=\"S\", delimiter=delim) # para csv\n if(encabezado):\n encabezado = todo[0,:]\n todo = todo[1:,:]\n else: \n encabezado = None\n \n clases = todo[:, i_clase]\n datos = np.delete(todo, i_clase, axis=1)\n print (\"Clases\")\n for k,v in Counter(clases).items(): print (k,\":\",v)\n return (datos, clases, encabezado) ", "_____no_output_____" ] ], [ [ "Ahora importando datos se hace muy simple en el futuro. Para datos de csv con cabezazo por ejemplo:", "_____no_output_____" ] ], [ [ "datos, clases, encabezado = lea_datos(\"datos_peña.csv\") # _ significa que no nos interesa este valor \nclases.shape", "Clases\nb'1' : 9\nb'0' : 14\n" ] ], [ [ "## Balanceando datos", "_____no_output_____" ], [ "Normalizacion de datos no es necesario para bosques aleatorios o arboles porque son invariantes al respeto de los magnitudes de variables. Lo unico que podría ser un problema es los clases son imbalanceados bajo nuestra percepción. Esto significa que a veces ni esperamos que nuestros datos van a ser balanceados y, en este caso, es permisible dejarlos sin balancear porque queremos que el clasificador incluye este probabilidad \"prior\" de la vida real. Si por ejemplo clasifico por ejemplo evaluaciones del vino, puede ser que lo encuentro normal que muchos vinos tienen una evaluacion promedia y solamente poco una muy buena. Si quiero que mi clasificador también tenga esta \"tendencía\" no tengo que balancear. \nEn otro caso, si diseño una prueba para una enfermeda y no quiero que mi clasificador hace asumpciones sobre el estado del paciente no basados en mis variables, mejor balanceo.\n\nScikit-learn no llega con una función para balancear datos entonces porque no lo hacemos nosotros?", "_____no_output_____" ] ], [ [ "def balance(datos, clases, estrategia=\"down\"):\n '''Balancea unos datos así que cada clase aparece en las mismas proporciones.\n Argumentos:\n datos - los datos. Filas son muestras y columnas variables.\n clases - las clases para cada muestra.\n estrategia - \"up\" para up-scaling y \"down\" para down-scaling'''\n \n import numpy as np\n from collections import Counter\n \n # Decidimos los nuevos numeros de muestras para cada clase\n conteos = Counter(clases)\n if estrategia==\"up\": muestras = max( conteos.values() )\n else: muestras = min( conteos.values() )\n \n datos_nuevo = np.array([]).reshape( 0, datos.shape[1] )\n clases_nuevo = []\n \n for c in conteos:\n c_i = np.where(clases==c)[0]\n new_i = np.random.choice(c_i, muestras, replace=(estrategia==\"up\") )\n datos_nuevo = np.append( datos_nuevo, datos[new_i,:], axis=0 )\n clases_nuevo = np.append( clases_nuevo, clases[new_i] )\n \n return (datos_nuevo, clases_nuevo)\n ", "_____no_output_____" ] ], [ [ "Y vamos a ver si funciona...", "_____no_output_____" ] ], [ [ "datos_b, clases_b = balance(datos, clases)\nprint Counter(clases_b)\nprint datos_b.shape", "_____no_output_____" ] ], [ [ "## Entrenando el bosque", "_____no_output_____" ], [ "Corriendo un bosque aleatorio para la clasificación es casi lo mismo como todos los otros clasificadores. Nada más importamos de una libreria diferente.", "_____no_output_____" ] ], [ [ "from sklearn.ensemble import RandomForestClassifier\n\nclf = RandomForestClassifier(oob_score=True)\nclf.fit(datos_b, clases_b)\nprint \"Exactitud estimado:\", clf.oob_score_", "_____no_output_____" ] ], [ [ "Ya sabemos que los hiper-parametros mas importantes para los bosques son el numero de arboles dado por `n_estimators` y la profundidad del arbol dado por `max_depth`. Porque son monotonos los podemos variar por separado. ", "_____no_output_____" ] ], [ [ "%matplotlib inline\nfrom sklearn.cross_validation import StratifiedKFold\nfrom sklearn.grid_search import GridSearchCV\nimport matplotlib.pyplot as plt\n\ncv = StratifiedKFold(y=clases, n_folds=4)\narboles_vals = np.arange(5,200,5)\nbusqueda = GridSearchCV(clf, param_grid=dict(n_estimators=arboles_vals), cv=cv)\nbusqueda.fit(datos, clases)\n\nprint 'Mejor numero de arboles=',busqueda.best_params_,',exactitud =',busqueda.best_score_\n\nscores = [x[1] for x in busqueda.grid_scores_]\n\nplt.plot(arboles_vals, scores)\nplt.xlabel('C')\nplt.ylabel('exactitud cv')\nplt.show()", "_____no_output_____" ] ], [ [ "Y para la profundidad:", "_____no_output_____" ] ], [ [ "prof_vals = np.arange(1,12)\nbusqueda = GridSearchCV(clf, param_grid=dict(max_depth=prof_vals), cv=cv)\nbusqueda.fit(datos, clases)\n\nprint 'Mejor profundidad=',busqueda.best_params_,',exactitud =',busqueda.best_score_\n\nscores = [x[1] for x in busqueda.grid_scores_]\nplt.plot(prof_vals, scores)\nplt.xlabel('profundidad maxima')\nplt.ylabel('exactitud cv')\nplt.show()", "_____no_output_____" ] ], [ [ "Viendo unas buenas valores podemos escoger un bosque con la menor numero de arboles y profundidad para una buena exactitud. También esta vez vamos a sacar las importancias de variables de una vez.", "_____no_output_____" ] ], [ [ "clf = RandomForestClassifier(n_estimators=101, oob_score=True) \nclf.fit(datos, clases)\nprint \"Exactitud estimada:\", clf.oob_score_\n\nprint cabeza\nprint clf.feature_importances_", "_____no_output_____" ] ], [ [ "## Predecir nuevas variables", "_____no_output_____" ] ], [ [ "datos_test, clases_test, _ = lea_datos(\"titanic_test.csv\")\nclases_pred = clf.predict(datos_test)\nprint \"Predicho:\", clases_pred\nprint \"Verdad: \", clases_test", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "Initialize estimator class", "_____no_output_____" ] ], [ [ "from __future__ import annotations\n\nfrom typing import NoReturn\n\nimport numpy as np\nimport pandas as pd\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.metrics import RocCurveDisplay, accuracy_score\n\nfrom IMLearn.base import BaseEstimator\n\nimport re\nfrom copy import copy\nfrom datetime import datetime\nfrom typing import NoReturn\n\nfrom sklearn.model_selection import GridSearchCV\nfrom sklearn.pipeline import Pipeline\nfrom sklearn.preprocessing import FunctionTransformer\n\nfrom IMLearn import BaseEstimator\nfrom challenge.agoda_cancellation_estimator import AgodaCancellationEstimator\n\n\n\nclass AgodaCancellationEstimator(BaseEstimator):\n def __init__(self, threshold: float = None) -> AgodaCancellationEstimator:\n super().__init__()\n self.__fit_model: RandomForestClassifier = None\n self.thresh = threshold\n\n def get_params(self, deep=False):\n return {'threshold': self.thresh}\n\n def set_params(self, threshold) -> AgodaCancellationEstimator:\n self.thresh = threshold\n\n return self\n\n def _fit(self, X: np.ndarray, y: np.ndarray) -> NoReturn:\n self.__fit_model = RandomForestClassifier(random_state=0).fit(X, y)\n\n def _predict(self, X: pd.DataFrame) -> np.ndarray:\n probs = self.__fit_model.predict_proba(X)[:, 1]\n return probs > self.thresh if self.thresh is not None else probs\n\n def _loss(self, X: np.ndarray, y: np.ndarray) -> float:\n pass\n\n def plot_roc_curve(self, X: np.ndarray, y: np.ndarray):\n RocCurveDisplay.from_estimator(self.__fit_model, X, y)\n\n def score(self, X: pd.DataFrame, y: pd.Series):\n return accuracy_score(y, self._predict(X))", "_____no_output_____" ] ], [ [ "Helper Functions", "_____no_output_____" ] ], [ [ "def read_data_file(path: str) -> pd.DataFrame:\n return pd.read_csv(path).drop_duplicates() \\\n .astype({'checkout_date': 'datetime64',\n 'checkin_date': 'datetime64',\n 'hotel_live_date': 'datetime64',\n 'booking_datetime': 'datetime64'})\n\n\ndef get_days_between_dates(dates1: pd.Series, dates2: pd.Series):\n return (dates1 - dates2).apply(lambda period: period.days)\n\n\ndef create_col_prob_mapper(col: str, mapper: dict):\n mapper = copy(mapper)\n\n def map_col_to_prob(df):\n df[col] = df[col].apply(mapper.get)\n\n return df\n\n return map_col_to_prob\n\n\ndef add_categorical_prep_to_pipe(train_features: pd.DataFrame, pipeline: Pipeline, cat_vars: list, one_hot=False,\n calc_probs=True) -> Pipeline:\n assert one_hot ^ calc_probs, \\\n 'Error: can only do either one-hot encoding or probability calculations, not neither/both!'\n # one-hot encoding\n if one_hot:\n # TODO - use sklearn OneHotEncoder\n pipeline.steps.append(('one-hot encoding',\n FunctionTransformer(lambda df: pd.get_dummies(df, columns=cat_vars))))\n\n # category probability preprocessing - make each category have its success percentage\n if calc_probs:\n for cat_var in cat_vars:\n map_cat_to_prob: dict = train_features.groupby(cat_var, dropna=False).labels.mean().to_dict()\n\n pipeline.steps.append((f'map {cat_var} to prob',\n FunctionTransformer(create_col_prob_mapper(cat_var, map_cat_to_prob))))\n\n return pipeline\n\n\ndef get_week_of_year(dates):\n return dates.apply(lambda d: d.weekofyear)\n\n\ndef get_booked_on_weekend(dates):\n return dates.apply(lambda d: d.day_of_week >= 4)\n\n\ndef get_weekend_holiday(in_date, out_date):\n return list(map(lambda d: (d[1] - d[0]).days <= 3 and d[0].dayofweek >= 4, zip(in_date, out_date)))\n\n\ndef get_local_holiday(col1, col2):\n return list(map(lambda x: x[0] == x[1], zip(col1, col2)))\n\n\ndef get_days_until_policy(policy_code: str) -> list:\n policies = policy_code.split('_')\n return [int(policy.split('D')[0]) if 'D' in policy else 0 for policy in policies]\n\n\ndef get_policy_cost(policy, stay_cost, stay_length, time_until_checkin):\n \"\"\"\n returns tuple of the format (max lost, min lost, part min lost)\n \"\"\"\n if policy == 'UNKNOWN':\n return 0, 0, 0\n nums = tuple(map(int, re.split('[a-zA-Z]', policy)[:-1]))\n if 'D' not in policy: # no show is suppressed\n return 0, 0, 0\n if 'N' in policy:\n nights_cost = stay_cost / stay_length * nums[0]\n min_cost = nights_cost if time_until_checkin <= nums[1] else 0\n return nights_cost, min_cost, min_cost / stay_cost\n elif 'P' in policy:\n nights_cost = stay_cost * nums[0] / 100\n min_cost = nights_cost if time_until_checkin <= nums[1] else 0\n return nights_cost, min_cost, min_cost / stay_cost\n else:\n raise Exception(\"Invalid Input\")\n\n\ndef get_money_lost_per_policy(features: pd.Series) -> list:\n policies = features.cancellation_policy_code.split('_')\n stay_cost = features.original_selling_amount\n stay_length = features.stay_length\n time_until_checkin = features.booking_to_arrival_time\n policy_cost = [get_policy_cost(policy, stay_cost, stay_length, time_until_checkin) for policy in policies]\n\n return list(map(list, zip(*policy_cost)))\n\n\ndef add_cancellation_policy_features(features: pd.DataFrame) -> pd.DataFrame:\n cancellation_policy = features.cancellation_policy_code\n features['n_policies'] = cancellation_policy.apply(lambda policy: len(policy.split('_')))\n days_until_policy = cancellation_policy.apply(get_days_until_policy)\n\n features['min_policy_days'] = days_until_policy.apply(min)\n features['max_policy_days'] = days_until_policy.apply(max)\n\n x = features.apply(get_money_lost_per_policy, axis='columns')\n features['max_policy_cost'], features['min_policy_cost'], features['part_min_policy_cost'] = list(\n map(list, zip(*x)))\n\n features['min_policy_cost'] = features['min_policy_cost'].apply(min)\n features['part_min_policy_cost'] = features['part_min_policy_cost'].apply(min)\n features['max_policy_cost'] = features['max_policy_cost'].apply(max)\n\n return features\n\n\ndef add_time_based_cols(df: pd.DataFrame) -> pd.DataFrame:\n df['stay_length'] = get_days_between_dates(df.checkout_date, df.checkin_date)\n df['time_registered_pre_book'] = get_days_between_dates(df.checkin_date, df.hotel_live_date)\n df['booking_to_arrival_time'] = get_days_between_dates(df.checkin_date, df.booking_datetime)\n df['checkin_week_of_year'] = get_week_of_year(df.checkin_date)\n df['booking_week_of_year'] = get_week_of_year(df.booking_datetime)\n df['booked_on_weekend'] = get_booked_on_weekend(df.booking_datetime)\n df['is_weekend_holiday'] = get_weekend_holiday(df.checkin_date, df.checkout_date)\n df['is_local_holiday'] = get_local_holiday(df.origin_country_code, df.hotel_country_code)\n\n return df\n", "_____no_output_____" ] ], [ [ "Define pipeline", "_____no_output_____" ] ], [ [ "def create_pipeline_from_data(filename: str):\n NONE_OUTPUT_COLUMNS = ['checkin_date',\n 'checkout_date',\n 'booking_datetime',\n 'hotel_live_date',\n 'hotel_country_code',\n 'origin_country_code',\n 'cancellation_policy_code']\n CATEGORICAL_COLUMNS = ['hotel_star_rating',\n 'guest_nationality_country_name',\n 'charge_option',\n 'accommadation_type_name',\n 'language',\n 'is_first_booking',\n 'customer_nationality',\n 'original_payment_currency',\n 'is_user_logged_in',\n ]\n RELEVANT_COLUMNS = ['no_of_adults',\n 'no_of_children',\n 'no_of_extra_bed',\n 'no_of_room',\n 'original_selling_amount'] + NONE_OUTPUT_COLUMNS + CATEGORICAL_COLUMNS\n features = read_data_file(filename)\n\n features['labels'] = features[\"cancellation_datetime\"].isna()\n\n pipeline_steps = [('columns selector', FunctionTransformer(lambda df: df[RELEVANT_COLUMNS])),\n ('add time based columns', FunctionTransformer(add_time_based_cols)),\n ('add cancellation policy features', FunctionTransformer(add_cancellation_policy_features))\n ]\n\n pipeline = Pipeline(pipeline_steps)\n pipeline = add_categorical_prep_to_pipe(features, pipeline, CATEGORICAL_COLUMNS)\n\n pipeline.steps.append( ('drop irrelevant columns', FunctionTransformer(lambda df: df.drop(NONE_OUTPUT_COLUMNS, axis='columns'))))\n\n return features.drop('labels', axis='columns'), features.labels, pipeline\n", "_____no_output_____" ] ], [ [ "Train, predict and export", "_____no_output_____" ] ], [ [ "def evaluate_and_export(estimator: BaseEstimator, X: pd.DataFrame, filename: str):\n preds = (~estimator.predict(X)).astype(int)\n pd.DataFrame(preds, columns=[\"predicted_values\"]).to_csv(filename, index=False)\n\n\ndef create_estimator_from_data(path=\"../datasets/agoda_cancellation_train.csv\", threshold: float = 0.47,\n optimize_threshold=False, debug=False) -> Pipeline:\n np.random.seed(0)\n\n # Load data\n raw_df, cancellation_labels, pipeline = create_pipeline_from_data(path)\n\n train_X = raw_df\n train_y = cancellation_labels\n train_X = pipeline.transform(train_X)\n\n # Fit model over data\n estimator = AgodaCancellationEstimator(threshold).fit(train_X, train_y)\n pipeline.steps.append(('estimator', estimator))\n\n return pipeline\n\ndef export_test_data(pipeline: Pipeline, path=\"../datasets/test_set_week_1.csv\") -> NoReturn:\n data = read_data_file(path)\n\n # Store model predictions over test set\n id1, id2, id3 = 209855253, 205843964, 212107536\n evaluate_and_export(pipeline, data, f\"{id1}_{id2}_{id3}.csv\")\n\npipeline = create_estimator_from_data()\nexport_test_data(pipeline)\n", "C:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:117: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:118: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:119: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:120: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:121: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:122: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:123: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:124: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:99: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:102: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:103: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:107: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:109: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:110: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:111: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:17: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:117: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:118: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:119: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:120: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:121: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:122: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:123: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:124: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:99: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:102: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:103: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:107: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:109: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:110: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:111: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\nC:\\conda\\envs\\iml.env\\lib\\site-packages\\ipykernel_launcher.py:17: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "# Let's keep our notebook clean, so it's a little more readable!\nimport warnings\nwarnings.filterwarnings('ignore')", "_____no_output_____" ], [ "%matplotlib inline", "_____no_output_____" ] ], [ [ "# Machine learning to predict age from rs-fmri\n\nThe goal is to extract data from several rs-fmri images, and use that data as features in a machine learning model. We will integrate what we've learned in the previous machine learning lecture to build an unbiased model and test it on a left out sample.\n\nWe're going to use a dataset that was prepared for this tutorial by [Elizabeth Dupre](https://elizabeth-dupre.com/#/), [Jake Vogel](https://github.com/illdopejake) and [Gael Varoquaux](http://gael-varoquaux.info/), by preprocessing [ds000228](https://openneuro.org/datasets/ds000228/versions/1.0.0) (from [Richardson et al. (2018)](https://dx.doi.org/10.1038%2Fs41467-018-03399-2)) through [fmriprep](https://github.com/poldracklab/fmriprep). They also created this tutorial and should be credited for it. ", "_____no_output_____" ], [ "## Load the data\n\n<img src=\"data/SampFeat.png\" alt=\"terms\" width=\"300\"/>\n", "_____no_output_____" ] ], [ [ "# change this to the location where you downloaded the data\nwdir = '/data/ml_tutorial/'\n", "_____no_output_____" ], [ "# Now fetch the data\n\nfrom glob import glob\nimport os\ndata = sorted(glob(os.path.join(wdir,'*.gz')))\nconfounds = sorted(glob(os.path.join(wdir,'*regressors.tsv')))", "_____no_output_____" ] ], [ [ "How many individual subjects do we have?", "_____no_output_____" ] ], [ [ "#len(data.func)\nlen(data)", "_____no_output_____" ] ], [ [ "## Extract features\n\n", "_____no_output_____" ], [ "In order to do our machine learning, we will need to extract feature from our rs-fmri images.\n\nSpecifically, we will extract signals from a brain parcellation and compute a correlation matrix, representing regional coactivation between regions.\n\nWe will practice on one subject first, then we'll extract data for all subjects", "_____no_output_____" ], [ "#### Retrieve the atlas for extracting features and an example subject", "_____no_output_____" ], [ "Since we're using rs-fmri data, it makes sense to use an atlas defined using rs-fmri data\n\nThis paper has many excellent insights about what kind of atlas to use for an rs-fmri machine learning task. See in particular Figure 5.\n\nhttps://www.sciencedirect.com/science/article/pii/S1053811919301594?via%3Dihub\n\nLet's use the MIST atlas, created here in Montreal using the BASC method. This atlas has multiple resolutions, for larger networks or finer-grained ROIs. Let's use a 64-ROI atlas to allow some detail, but to ultimately keep our connectivity matrices manageable\n\nHere is a link to the MIST paper: https://mniopenresearch.org/articles/1-3", "_____no_output_____" ] ], [ [ "from nilearn import datasets\n\nparcellations = datasets.fetch_atlas_basc_multiscale_2015(version='sym')\natlas_filename = parcellations.scale064\n\n\nprint('Atlas ROIs are located in nifti image (4D) at: %s' %\n atlas_filename)", "_____no_output_____" ] ], [ [ "Let's have a look at that atlas", "_____no_output_____" ] ], [ [ "from nilearn import plotting\n\nplotting.plot_roi(atlas_filename, draw_cross=False)", "_____no_output_____" ] ], [ [ "Great, let's load an example 4D fmri time-series for one subject", "_____no_output_____" ] ], [ [ "fmri_filenames = data[0]\nprint(fmri_filenames)", "_____no_output_____" ] ], [ [ "Let's have a look at the image! Because it is a 4D image, we can only look at one slice at a time. Or, better yet, let's look at an average image!", "_____no_output_____" ] ], [ [ "from nilearn import image \n\naveraged_Img = image.mean_img(image.mean_img(fmri_filenames))\nplotting.plot_stat_map(averaged_Img)", "_____no_output_____" ] ], [ [ "#### Extract signals on a parcellation defined by labels\nUsing the NiftiLabelsMasker\n\nSo we've loaded our atlas and 4D data for a single subject. Let's practice extracting features!\n", "_____no_output_____" ] ], [ [ "from nilearn.input_data import NiftiLabelsMasker\n\nmasker = NiftiLabelsMasker(labels_img=atlas_filename, standardize=True, \n memory='nilearn_cache', verbose=1)\n\n# Here we go from nifti files to the signal time series in a numpy\n# array. Note how we give confounds to be regressed out during signal\n# extraction\nconf = confounds[0]\ntime_series = masker.fit_transform(fmri_filenames, confounds=conf)", "_____no_output_____" ] ], [ [ "So what did we just create here?", "_____no_output_____" ] ], [ [ "type(time_series)", "_____no_output_____" ], [ "time_series.shape", "_____no_output_____" ] ], [ [ "What are these \"confounds\" and how are they used? ", "_____no_output_____" ] ], [ [ "import pandas\nconf_df = pandas.read_table(conf)\nconf_df.head()", "_____no_output_____" ], [ "conf_df.shape", "_____no_output_____" ] ], [ [ "#### Compute and display a correlation matrix\n", "_____no_output_____" ] ], [ [ "from nilearn.connectome import ConnectivityMeasure\n\ncorrelation_measure = ConnectivityMeasure(kind='correlation')\ncorrelation_matrix = correlation_measure.fit_transform([time_series])[0]\ncorrelation_matrix.shape", "_____no_output_____" ] ], [ [ "Plot the correlation matrix", "_____no_output_____" ] ], [ [ "import numpy as np\n# Mask the main diagonal for visualization:\nnp.fill_diagonal(correlation_matrix, 0)\n\n# The labels we have start with the background (0), hence we skip the\n# first label\nplotting.plot_matrix(correlation_matrix, figure=(10, 8), \n labels=range(time_series.shape[-1]),\n vmax=0.8, vmin=-0.8, reorder=False)\n\n# matrices are ordered for block-like representation", "_____no_output_____" ] ], [ [ "#### Extract features from the whole dataset\n\nHere, we are going to use a for loop to iterate through each image and use the same techniques we learned above to extract rs-fmri connectivity features from every subject.\n\n", "_____no_output_____" ] ], [ [ "# Here is a really simple for loop\n\nfor i in range(10):\n print('the number is', i)", "_____no_output_____" ], [ "container = []\nfor i in range(10):\n container.append(i)\n\ncontainer", "_____no_output_____" ] ], [ [ "Now lets construct a more complicated loop to do what we want", "_____no_output_____" ], [ "First we do some things we don't need to do in the loop. Let's reload our atlas, and re-initiate our masker and correlation_measure", "_____no_output_____" ] ], [ [ "from nilearn.input_data import NiftiLabelsMasker\nfrom nilearn.connectome import ConnectivityMeasure\nfrom nilearn import datasets\n\n# load atlas\nmultiscale = datasets.fetch_atlas_basc_multiscale_2015()\natlas_filename = multiscale.scale064\n\n# initialize masker (change verbosity)\nmasker = NiftiLabelsMasker(labels_img=atlas_filename, standardize=True, \n memory='nilearn_cache', verbose=0)\n\n# initialize correlation measure, set to vectorize\ncorrelation_measure = ConnectivityMeasure(kind='correlation', vectorize=True,\n discard_diagonal=True)", "_____no_output_____" ] ], [ [ "Okay -- now that we have that taken care of, let's run our big loop!", "_____no_output_____" ], [ "**NOTE**: On a laptop, this might a few minutes.", "_____no_output_____" ] ], [ [ "all_features = [] # here is where we will put the data (a container)\n\nfor i,sub in enumerate(data):\n # extract the timeseries from the ROIs in the atlas\n time_series = masker.fit_transform(sub, confounds=confounds[i])\n # create a region x region correlation matrix\n correlation_matrix = correlation_measure.fit_transform([time_series])[0]\n # add to our container\n all_features.append(correlation_matrix)\n # keep track of status\n print('finished %s of %s'%(i+1,len(data)))", "_____no_output_____" ], [ "# Let's save the data to disk\nimport numpy as np\n\nnp.savez_compressed('MAIN_BASC064_subsamp_features',a = all_features)", "_____no_output_____" ] ], [ [ "In case you do not want to run the full loop on your computer, you can load the output of the loop here!", "_____no_output_____" ] ], [ [ "feat_file = 'MAIN_BASC064_subsamp_features.npz'\nX_features = np.load(feat_file)['a']", "_____no_output_____" ], [ "X_features.shape", "_____no_output_____" ] ], [ [ "<img src=\"data/SampFeat.png\" alt=\"terms\" width=\"300\"/>", "_____no_output_____" ], [ "Okay so we've got our features.", "_____no_output_____" ], [ "We can visualize our feature matrix", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\n\nplt.imshow(X_features, aspect='auto')\nplt.colorbar()\nplt.title('feature matrix')\nplt.xlabel('features')\nplt.ylabel('subjects')", "_____no_output_____" ] ], [ [ "## Get Y (our target) and assess its distribution", "_____no_output_____" ] ], [ [ "# Let's load the phenotype data\n\npheno_path = os.path.join(wdir, 'participants.tsv')", "_____no_output_____" ], [ "import pandas\n\npheno = pandas.read_csv(pheno_path, sep='\\t').sort_values('participant_id')\npheno.head()", "_____no_output_____" ] ], [ [ "Looks like there is a column labeling age. Let's capture it in a variable", "_____no_output_____" ] ], [ [ "y_age = pheno['Age']", "_____no_output_____" ] ], [ [ "Maybe we should have a look at the distribution of our target variable", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nimport seaborn as sns\nsns.distplot(y_age)", "_____no_output_____" ] ], [ [ "## Prepare data for machine learning\n\nHere, we will define a \"training sample\" where we can play around with our models. We will also set aside a \"validation\" sample that we will not touch until the end", "_____no_output_____" ], [ "We want to be sure that our training and test sample are matched! We can do that with a \"stratified split\". This dataset has a variable indicating AgeGroup. We can use that to make sure our training and testing sets are balanced!", "_____no_output_____" ] ], [ [ "age_class = pheno['AgeGroup']\nage_class.value_counts()", "_____no_output_____" ], [ "from sklearn.model_selection import train_test_split\n\n# Split the sample to training/validation with a 60/40 ratio, and \n# stratify by age class, and also shuffle the data.\n\nX_train, X_val, y_train, y_val = train_test_split(\n X_features, # x\n y_age, # y\n test_size = 0.4, # 60%/40% split \n shuffle = True, # shuffle dataset\n # before splitting\n stratify = age_class, # keep\n # distribution\n # of ageclass\n # consistent\n # betw. train\n # & test sets.\n random_state = 123 # same shuffle each\n # time\n )\n\n# print the size of our training and test groups\nprint('training:', len(X_train),\n 'testing:', len(X_val))", "_____no_output_____" ] ], [ [ "Let's visualize the distributions to be sure they are matched", "_____no_output_____" ] ], [ [ "sns.distplot(y_train)\nsns.distplot(y_val)", "_____no_output_____" ] ], [ [ "## Run your first model!\n\nMachine learning can get pretty fancy pretty quickly. We'll start with a fairly standard regression model called a Support Vector Regressor (SVR). \n\nWhile this may seem unambitious, simple models can be very robust. And we probably don't have enough data to create more complex models (but we can try later).\n\nFor more information, see this excellent resource:\nhttps://hal.inria.fr/hal-01824205", "_____no_output_____" ], [ "Let's fit our first model!", "_____no_output_____" ] ], [ [ "from sklearn.svm import SVR\nl_svr = SVR(kernel='linear') # define the model\n\nl_svr.fit(X_train, y_train) # fit the model", "_____no_output_____" ] ], [ [ "Well... that was easy. Let's see how well the model learned the data!\n\n\n<img src=\"data/modval.png\" alt=\"terms\" width=\"800\"/>", "_____no_output_____" ] ], [ [ "\n# predict the training data based on the model\ny_pred = l_svr.predict(X_train) \n\n# caluclate the model accuracy\nacc = l_svr.score(X_train, y_train) \n", "_____no_output_____" ] ], [ [ "Let's view our results and plot them all at once!", "_____no_output_____" ] ], [ [ "# print results\nprint('accuracy (R2)', acc)\n\nsns.regplot(y_pred,y_train)\nplt.xlabel('Predicted Age')", "_____no_output_____" ] ], [ [ "HOLY COW! Machine learning is amazing!!! Almost a perfect fit!\n\n...which means there's something wrong. What's the problem here?", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split\n\n# Split the sample to training/test with a 75/25 ratio, and \n# stratify by age class, and also shuffle the data.\n\nage_class2 = pheno.loc[y_train.index,'AgeGroup']\n\nX_train2, X_test, y_train2, y_test = train_test_split(\n X_train, # x\n y_train, # y\n test_size = 0.25, # 75%/25% split \n shuffle = True, # shuffle dataset\n # before splitting\n stratify = age_class2, # keep\n # distribution\n # of ageclass\n # consistent\n # betw. train\n # & test sets.\n random_state = 123 # same shuffle each\n # time\n )\n\n# print the size of our training and test groups\nprint('training:', len(X_train2),\n 'testing:', len(X_test))", "_____no_output_____" ], [ "from sklearn.metrics import mean_absolute_error\n\n# fit model just to training data\nl_svr.fit(X_train2,y_train2)\n\n# predict the *test* data based on the model trained on X_train2\ny_pred = l_svr.predict(X_test) \n\n# caluclate the model accuracy\nacc = l_svr.score(X_test, y_test) \nmae = mean_absolute_error(y_true=y_test,y_pred=y_pred)", "_____no_output_____" ], [ "# print results\nprint('accuracy (R2) = ', acc)\nprint('MAE = ',mae)\n\nsns.regplot(y_pred,y_test)\nplt.xlabel('Predicted Age')", "_____no_output_____" ] ], [ [ "Not perfect, but as predicting with unseen data goes, not too bad! Especially with a training sample of \"only\" 69 subjects. But we can do better! \n\nFor example, we can increase the size our training set while simultaneously reducing bias by instead using 10-fold cross-validation\n\n<img src=\"data/KCV.png\" alt=\"terms\" width=\"500\"/>", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_predict, cross_val_score\n\n# predict\ny_pred = cross_val_predict(l_svr, X_train, y_train, cv=10)\n# scores\nacc = cross_val_score(l_svr, X_train, y_train, cv=10)\nmae = cross_val_score(l_svr, X_train, y_train, cv=10, scoring='neg_mean_absolute_error')", "_____no_output_____" ] ], [ [ "We can look at the accuracy of the predictions for each fold of the cross-validation", "_____no_output_____" ] ], [ [ "for i in range(10):\n print('Fold {} -- Acc = {}, MAE = {}'.format(i, acc[i],-mae[i]))", "_____no_output_____" ] ], [ [ "We can also look at the overall accuracy of the model", "_____no_output_____" ] ], [ [ "from sklearn.metrics import r2_score\n\noverall_acc = r2_score(y_train, y_pred)\noverall_mae = mean_absolute_error(y_train,y_pred)\nprint('R2:',overall_acc)\nprint('MAE:',overall_mae)\n\nsns.regplot(y_pred, y_train)\nplt.xlabel('Predicted Age')", "_____no_output_____" ] ], [ [ "Not too bad at all! But more importantly, this is a more accurate estimation of our model's predictive efficacy. Our sample size is larger and this is based on several rounds of prediction of unseen data. \n\nFor example, we can now see that the effect is being driven by the model's successful parsing of adults vs. children, but is not performing so well within the adult or children group. This was not evident during our previous iteration of the model", "_____no_output_____" ], [ "## Tweak your model\n\nIt's very important to learn when and where its appropriate to \"tweak\" your model.\n\nSince we have done all of the previous analysis in our training data, it's fine to try out different models. But we **absolutely cannot** \"test\" it on our left out data. If we do, we are in great danger of overfitting.\n\nIt is not uncommon to try other models, or tweak hyperparameters. In this case, due to our relatively small sample size, we are probably not powered sufficiently to do so, and we would once again risk overfitting. However, for the sake of demonstration, we will do some tweaking. \n\n<img src=\"data/KCV2.png\" alt=\"terms\" width=\"500\"/>\n", "_____no_output_____" ], [ "We will try a few different examples:\n* Normalizing our target data\n* Tweaking our hyperparameters\n* Trying a more complicated model\n* Feature selection", "_____no_output_____" ], [ "#### Normalize the target data", "_____no_output_____" ] ], [ [ "# Create a log transformer function and log transform Y (age)\n\nfrom sklearn.preprocessing import FunctionTransformer\n\nlog_transformer = FunctionTransformer(func = np.log, validate=True)\nlog_transformer.fit(y_train.values.reshape(-1,1))\ny_train_log = log_transformer.transform(y_train.values.reshape(-1,1))[:,0]\n\nsns.distplot(y_train_log)\nplt.title(\"Log-Transformed Age\")", "_____no_output_____" ] ], [ [ "Now let's go ahead and cross-validate our model once again with this new log-transformed target", "_____no_output_____" ] ], [ [ "# predict\ny_pred = cross_val_predict(l_svr, X_train, y_train_log, cv=10)\n\n# scores\nacc = r2_score(y_train_log, y_pred)\nmae = mean_absolute_error(y_train_log,y_pred)\n\nprint('R2:',acc)\nprint('MAE:',mae)\n\nsns.regplot(y_pred, y_train_log)\nplt.xlabel('Predicted Log Age')\nplt.ylabel('Log Age')", "_____no_output_____" ] ], [ [ "Seems like a definite improvement, right? I think we can agree on that. \n\nBut we can't forget about interpretability? The MAE is much less interpretable now.", "_____no_output_____" ], [ "#### Tweak the hyperparameters\n\nMany machine learning algorithms have hyperparameters that can be \"tuned\" to optimize model fitting. Careful parameter tuning can really improve a model, but haphazard tuning will often lead to overfitting.\n\nOur SVR model has multiple hyperparameters. Let's explore some approaches for tuning them", "_____no_output_____" ] ], [ [ "SVR?", "_____no_output_____" ] ], [ [ "One way is to plot a \"Validation Curve\" -- this will let us view changes in training and validation accuracy of a model as we shift its hyperparameters. We can do this easily with sklearn.", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import validation_curve\n\nC_range = 10. ** np.arange(-3, 8) # A range of different values for C\n\ntrain_scores, valid_scores = validation_curve(l_svr, X_train, y_train_log, \n param_name= \"C\",\n param_range = C_range,\n cv=10)\n", "_____no_output_____" ], [ "# A bit of pandas magic to prepare the data for a seaborn plot\n\ntScores = pandas.DataFrame(train_scores).stack().reset_index()\ntScores.columns = ['C','Fold','Score']\ntScores.loc[:,'Type'] = ['Train' for x in range(len(tScores))]\n\nvScores = pandas.DataFrame(valid_scores).stack().reset_index()\nvScores.columns = ['C','Fold','Score']\nvScores.loc[:,'Type'] = ['Validate' for x in range(len(vScores))]\n\nValCurves = pandas.concat([tScores,vScores]).reset_index(drop=True)\nValCurves.head()", "_____no_output_____" ], [ "# And plot!\n# g = sns.lineplot(x='C',y='Score',hue='Type',data=ValCurves)\n# g.set_xticks(range(10))\n# g.set_xticklabels(C_range, rotation=90)\n\ng = sns.factorplot(x='C',y='Score',hue='Type',data=ValCurves)\nplt.xticks(range(10))\ng.set_xticklabels(C_range, rotation=90)", "_____no_output_____" ] ], [ [ "It looks like accuracy is better for higher values of C, and plateaus somewhere between 0.1 and 1. The default setting is C=1, so it looks like we can't really improve by changing C.", "_____no_output_____" ], [ "But our SVR model actually has two hyperparameters, C and epsilon. Perhaps there is an optimal combination of settings for these two parameters.\n\nWe can explore that somewhat quickly with a grid search, which is once again easily achieved with sklearn. Because we are fitting the model multiple times witih cross-validation, this will take some time", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import GridSearchCV\n\nC_range = 10. ** np.arange(-3, 8)\nepsilon_range = 10. ** np.arange(-3, 8)\n\nparam_grid = dict(epsilon=epsilon_range, C=C_range)\n\ngrid = GridSearchCV(l_svr, param_grid=param_grid, cv=10)\n\ngrid.fit(X_train, y_train_log)", "_____no_output_____" ] ], [ [ "Now that the grid search has completed, let's find out what was the \"best\" parameter combination", "_____no_output_____" ] ], [ [ "print(grid.best_params_)", "_____no_output_____" ] ], [ [ "And if redo our cross-validation with this parameter set?", "_____no_output_____" ] ], [ [ "y_pred = cross_val_predict(SVR(kernel='linear',C=0.10,epsilon=0.10, gamma='auto'), \n X_train, y_train_log, cv=10)\n\n# scores\nacc = r2_score(y_train_log, y_pred)\nmae = mean_absolute_error(y_train_log,y_pred)\n\nprint('R2:',acc)\nprint('MAE:',mae)\n\nsns.regplot(y_pred, y_train_log)\nplt.xlabel('Predicted Log Age')\nplt.ylabel('Log Age')", "_____no_output_____" ] ], [ [ "Perhaps unsurprisingly, the model fit is actually exactly the same as what we had with our defaults. There's a reason they are defaults ;-)\n\nGrid search can be a powerful and useful tool. But can you think of a way that, if not properly utilized, it could lead to overfitting?", "_____no_output_____" ], [ "You can find a nice set of tutorials with links to very helpful content regarding how to tune hyperparameters and while be aware of over- and under-fitting here:\n\nhttps://scikit-learn.org/stable/modules/learning_curve.html", "_____no_output_____" ], [ "#### Trying a more complicated model", "_____no_output_____" ], [ "In principle, there is no real reason to do this. Perhaps one could make an argument for quadratic relationship with age, but we probably don't have enough subjects to learn a complicated non-linear model. But for the sake of demonstration, we can give it a shot.\n\nWe'll use a validation curve to see the result of our model if, instead of fitting a linear model, we instead try to the fit a 2nd, 3rd, ... 8th order polynomial.", "_____no_output_____" ] ], [ [ "validation_curve?", "_____no_output_____" ], [ "from sklearn.model_selection import validation_curve\n\ndegree_range = list(range(1,8)) # A range of different values for C\n\ntrain_scores, valid_scores = validation_curve(SVR(kernel='poly',\n gamma='scale'\n ), \n X=X_train, y=y_train_log, \n param_name= \"degree\",\n param_range = degree_range,\n cv=10)\n", "_____no_output_____" ], [ "# A bit of pandas magic to prepare the data for a seaborn plot\n\ntScores = pandas.DataFrame(train_scores).stack().reset_index()\ntScores.columns = ['Degree','Fold','Score']\ntScores.loc[:,'Type'] = ['Train' for x in range(len(tScores))]\n\nvScores = pandas.DataFrame(valid_scores).stack().reset_index()\nvScores.columns = ['Degree','Fold','Score']\nvScores.loc[:,'Type'] = ['Validate' for x in range(len(vScores))]\n\nValCurves = pandas.concat([tScores,vScores]).reset_index(drop=True)\nValCurves.head()", "_____no_output_____" ], [ "# And plot!\n\n# g = sns.lineplot(x='Degree',y='Score',hue='Type',data=ValCurves)\n# g.set_xticks(range(10))\n# g.set_xticklabels(degree_range, rotation=90)\n\ng = sns.factorplot(x='Degree',y='Score',hue='Type',data=ValCurves)\nplt.xticks(range(10))\ng.set_xticklabels(degree_range, rotation=90)", "_____no_output_____" ] ], [ [ "It appears that we cannot improve our model by increasing the complexity of the fit. If one looked only at the training data, one might surmise that a 2nd order fit could be a slightly better model. But that improvement does not generalize to the validation data.", "_____no_output_____" ] ], [ [ "# y_pred = cross_val_predict(SVR(kernel='rbf', gamma='scale'), X_train, y_train_log, cv=10)\n\n# # scores\n# acc = r2_score(y_train_log, y_pred)\n# mae = mean_absolute_error(y_train_log,y_pred)\n\n# print('R2:',acc)\n# print('MAE:',mae)\n\n# sns.regplot(y_pred, y_train_log)\n# plt.xlabel('Predicted Log Age')\n# plt.ylabel('Log Age')", "_____no_output_____" ] ], [ [ "#### Feature selection\n\nRight now, we have 2016 features. Are all of those really going to contribute to the model stably? \n\nIntuitively, models tend to perform better when there are fewer, more important features than when there are many, less imortant features. The tough part is figuring out which features are useful or important.\n\nHere will quickly try a basic feature seclection strategy\n\n<img src=\"data/FeatSel.png\" alt=\"terms\" width=\"400\"/>", "_____no_output_____" ], [ "The SelectPercentile() function will select the top X% of features based on univariate tests. This is a way of identifying theoretically more useful features. But remember, significance != prediction! \n\nWe are also in danger of overfitting here. For starters, if we want to test this with 10-fold cross-validation, we will need to do a separate feature selection within each fold! That means we'll need to do the cross-validation manually instead of using cross_val_predict(). ", "_____no_output_____" ] ], [ [ "from sklearn.feature_selection import SelectPercentile, f_regression\nfrom sklearn.model_selection import KFold\nfrom sklearn.pipeline import Pipeline\n\n# Build a tiny pipeline that does feature selection (top 20% of features), \n# and then prediction with our linear svr model.\nmodel = Pipeline([\n ('feature_selection',SelectPercentile(f_regression,percentile=20)),\n ('prediction', l_svr)\n ])\n\ny_pred = [] # a container to catch the predictions from each fold\ny_index = [] # just in case, the index for each prediciton\n\n# First we create 10 splits of the data\nskf = KFold(n_splits=10, shuffle=True, random_state=123)\n\n# For each split, assemble the train and test samples \nfor tr_ind, te_ind in skf.split(X_train):\n X_tr = X_train[tr_ind]\n y_tr = y_train_log[tr_ind]\n X_te = X_train[te_ind]\n y_index += list(te_ind) # store the index of samples to predict\n \n # and run our pipeline \n model.fit(X_tr, y_tr) # fit the data to the model using our mini pipeline\n predictions = model.predict(X_te).tolist() # get the predictions for this fold\n y_pred += predictions # add them to the list of predictions\n\n\n ", "_____no_output_____" ] ], [ [ "Alrighty, let's see if only using the top 20% of features improves the model at all...", "_____no_output_____" ] ], [ [ "acc = r2_score(y_train_log[y_index], y_pred)\nmae = mean_absolute_error(y_train_log[y_index],y_pred)\n\nprint('R2:',acc)\nprint('MAE:',mae)\n\nsns.regplot(np.array(y_pred), y_train_log[y_index])\nplt.xlabel('Predicted Log Age')\nplt.ylabel('Log Age')\n", "_____no_output_____" ] ], [ [ "Nope, in fact it got a bit worse. It seems we're getting \"value at the margins\" so to speak. This is a very good example of how significance != prediction, as demonstrated in this figure from Bzdok et al., 2018 *bioRxiv*\n\n", "_____no_output_____" ], [ "See here for an explanation of different feature selection options and how to implement them in sklearn: https://scikit-learn.org/stable/modules/feature_selection.html\n\nAnd here is a thoughtful tutorial covering feature selection for novel machine learners: https://www.datacamp.com/community/tutorials/feature-selection-python", "_____no_output_____" ], [ "So there you have it. We've tried many different strategies, but most of our \"tweaks\" haven't really lead to improvements in the model. This is not always the case, but it is not uncommon. Can you think of some reasons why?\n\nMoving on to our validation data, we probably should just stick to a basic model, though predicting log age might be a good idea!", "_____no_output_____" ], [ "## Can our model predict age in completely un-seen data?\nNow that we've fit a model we think has possibly learned how to decode age based on rs-fmri signal, let's put it to the test. We will train our model on all of the training data, and try to predict the age of the subjects we left out at the beginning of this section.", "_____no_output_____" ], [ "Because we performed a log transformation on our training data, we will need to transform our testing data using the *same information!* But that's easy because we stored our transformation in an object!\n", "_____no_output_____" ] ], [ [ "# Notice how we use the Scaler that was fit to X_train and apply to X_test,\n# rather than creating a new Scaler for X_test\ny_val_log = log_transformer.transform(y_val.values.reshape(-1,1))[:,0]", "_____no_output_____" ] ], [ [ "And now for the moment of truth! \n\nNo cross-validation needed here. We simply fit the model with the training data and use it to predict the testing data\n\nI'm so nervous. Let's just do it all in one cell", "_____no_output_____" ] ], [ [ "l_svr.fit(X_train, y_train_log) # fit to training data\ny_pred = l_svr.predict(X_val) # classify age class using testing data\nacc = l_svr.score(X_val, y_val_log) # get accuracy (r2)\nmae = mean_absolute_error(y_val_log, y_pred) # get mae\n\n# print results\nprint('accuracy (r2) =', acc)\nprint('mae = ',mae)\n\n# plot results\nsns.regplot(y_pred, y_val_log)\nplt.xlabel('Predicted Log Age')\nplt.ylabel('Log Age')", "_____no_output_____" ] ], [ [ "***Wow!!*** Congratulations. You just trained a machine learning model that used real rs-fmri data to predict the age of real humans.\n\nThe proper thing to do at this point would be to repeat the train-validation split multiple times. This will ensure the results are not specific to this validation set, and will give you some confidence intervals around your results.\n\nAs an assignment, you can give that a try below. Create 10 different splits of the entire dataset, fit the model and get your predictions. Then, plot the range of predictions.\n", "_____no_output_____" ] ], [ [ "# SPACE FOR YOUR ASSIGNMENT\n\n\n", "_____no_output_____" ] ], [ [ "So, it seems like something in this data does seem to be systematically related to age ... but what? ", "_____no_output_____" ], [ "#### Interpreting model feature importances\nInterpreting the feature importances of a machine learning model is a real can of worms. This is an area of active research. Unfortunately, it's hard to trust the feature importance of some models. \n\nYou can find a whole tutorial on this subject here:\nhttp://gael-varoquaux.info/interpreting_ml_tuto/index.html\n\nFor now, we'll just eschew better judgement and take a look at our feature importances. While we can't ascribe any biological relevance to the features, it can still be helpful to know what the model is using to make its predictions. This is a good way to, for example, establish whether your model is actually learning based on a confound! Could you think of some examples?", "_____no_output_____" ], [ "We can access the feature importances (weights) used my the model", "_____no_output_____" ] ], [ [ "l_svr.coef_", "_____no_output_____" ] ], [ [ "lets plot these weights to see their distribution better", "_____no_output_____" ] ], [ [ "plt.bar(range(l_svr.coef_.shape[-1]),l_svr.coef_[0])\nplt.title('feature importances')\nplt.xlabel('feature')\nplt.ylabel('weight')", "_____no_output_____" ] ], [ [ "Or perhaps it will be easier to visualize this information as a matrix similar to the one we started with\n\nWe can use the correlation measure from before to perform an inverse transform", "_____no_output_____" ] ], [ [ "correlation_measure.inverse_transform(l_svr.coef_).shape", "_____no_output_____" ], [ "from nilearn import plotting\n\nfeat_exp_matrix = correlation_measure.inverse_transform(l_svr.coef_)[0]\n\nplotting.plot_matrix(feat_exp_matrix, figure=(10, 8), \n labels=range(feat_exp_matrix.shape[0]),\n reorder=False,\n tri='lower')", "_____no_output_____" ] ], [ [ "Let's see if we can throw those features onto an actual brain.\n\nFirst, we'll need to gather the coordinates of each ROI of our atlas", "_____no_output_____" ] ], [ [ "coords = plotting.find_parcellation_cut_coords(atlas_filename)", "_____no_output_____" ] ], [ [ "And now we can use our feature matrix and the wonders of nilearn to create a connectome map where each node is an ROI, and each connection is weighted by the importance of the feature to the model", "_____no_output_____" ] ], [ [ "plotting.plot_connectome(feat_exp_matrix, coords, colorbar=True)", "_____no_output_____" ] ], [ [ "Whoa!! That's...a lot to process. Maybe let's threshold the edges so that only the most important connections are visualized", "_____no_output_____" ] ], [ [ "plotting.plot_connectome(feat_exp_matrix, coords, colorbar=True, edge_threshold=0.035)", "_____no_output_____" ] ], [ [ "That's definitely an improvement, but it's still a bit hard to see what's going on.\nNilearn has a new feature that let's use view this data interactively!", "_____no_output_____" ] ], [ [ "plotting.view_connectome(feat_exp_matrix, coords, threshold='98%')", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import cerespp\nimport glob", "_____no_output_____" ], [ "files = glob.glob('HD10700*.fits')\nfiles", "_____no_output_____" ], [ "lines = ['CaHK', 'Ha', 'HeI', 'NaID1D2']\ncerespp.line_plot_from_file(files[0], lines, './', 'HD10700')", "Correcting spectrum to rest frame. (This could take a while)\nMerging echelle spectra.\nPlotting line CaHK\nPlotting line Ha\nPlotting line HeI\nPlotting line NaID1D2\n" ], [ "cerespp.get_activities(files, './HD10700_activities.dat')", "100%|██████████| 2/2 [06:15<00:00, 187.67s/it]\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code" ] ]

[ "CC0-1.0" ]

[ "CC0-1.0" ]

[ "CC0-1.0" ]

[ [ [ "<!--NAVIGATION-->\n< [Errors and Exceptions](09-Errors-and-Exceptions.ipynb) | [Contents](Index.ipynb) | [List Comprehensions](11-List-Comprehensions.ipynb) >", "_____no_output_____" ], [ "# Iterators", "_____no_output_____" ], [ "Often an important piece of data analysis is repeating a similar calculation, over and over, in an automated fashion.\nFor example, you may have a table of a names that you'd like to split into first and last, or perhaps of dates that you'd like to convert to some standard format.\nOne of Python's answers to this is the *iterator* syntax.\nWe've seen this already with the ``range`` iterator:", "_____no_output_____" ] ], [ [ "for i in range(10):\n print(i, end=' ')", "0 1 2 3 4 5 6 7 8 9 " ] ], [ [ "Here we're going to dig a bit deeper.\nIt turns out that in Python 3, ``range`` is not a list, but is something called an *iterator*, and learning how it works is key to understanding a wide class of very useful Python functionality.", "_____no_output_____" ], [ "## Iterating over lists\nIterators are perhaps most easily understood in the concrete case of iterating through a list.\nConsider the following:", "_____no_output_____" ] ], [ [ "for value in [2, 4, 6, 8, 10]:\n # do some operation\n print(value + 1, end=' ')", "3 5 7 9 11 " ] ], [ [ "The familiar \"``for x in y``\" syntax allows us to repeat some operation for each value in the list.\nThe fact that the syntax of the code is so close to its English description (\"*for [each] value in [the] list*\") is just one of the syntactic choices that makes Python such an intuitive language to learn and use.\n\nBut the face-value behavior is not what's *really* happening.\nWhen you write something like \"``for val in L``\", the Python interpreter checks whether it has an *iterator* interface, which you can check yourself with the built-in ``iter`` function:", "_____no_output_____" ] ], [ [ "iter([2, 4, 6, 8, 10])", "_____no_output_____" ] ], [ [ "It is this iterator object that provides the functionality required by the ``for`` loop.\nThe ``iter`` object is a container that gives you access to the next object for as long as it's valid, which can be seen with the built-in function ``next``:", "_____no_output_____" ] ], [ [ "I = iter([2, 4, 6, 8, 10])", "_____no_output_____" ], [ "print(next(I))", "2\n" ], [ "print(next(I))", "4\n" ], [ "print(next(I))", "6\n" ] ], [ [ "What is the purpose of this level of indirection?\nWell, it turns out this is incredibly useful, because it allows Python to treat things as lists that are *not actually lists*.", "_____no_output_____" ], [ "## ``range()``: A List Is Not Always a List\nPerhaps the most common example of this indirect iteration is the ``range()`` function in Python 3 (named ``xrange()`` in Python 2), which returns not a list, but a special ``range()`` object:", "_____no_output_____" ] ], [ [ "range(10)", "_____no_output_____" ] ], [ [ "``range``, like a list, exposes an iterator:", "_____no_output_____" ] ], [ [ "iter(range(10))", "_____no_output_____" ] ], [ [ "So Python knows to treat it *as if* it's a list:", "_____no_output_____" ] ], [ [ "for i in range(10):\n print(i, end=' ')", "0 1 2 3 4 5 6 7 8 9 " ] ], [ [ "The benefit of the iterator indirection is that *the full list is never explicitly created!*\nWe can see this by doing a range calculation that would overwhelm our system memory if we actually instantiated it (note that in Python 2, ``range`` creates a list, so running the following will not lead to good things!):", "_____no_output_____" ] ], [ [ "N = 10 ** 12\nfor i in range(N):\n if i >= 10: break\n print(i, end=', ')", "0, 1, 2, 3, 4, 5, 6, 7, 8, 9, " ] ], [ [ "If ``range`` were to actually create that list of one trillion values, it would occupy tens of terabytes of machine memory: a waste, given the fact that we're ignoring all but the first 10 values!\n\nIn fact, there's no reason that iterators ever have to end at all!\nPython's ``itertools`` library contains a ``count`` function that acts as an infinite range:", "_____no_output_____" ] ], [ [ "from itertools import count\n\nfor i in count():\n if i >= 10:\n break\n print(i, end=', ')", "0, 1, 2, 3, 4, 5, 6, 7, 8, 9, " ] ], [ [ "Had we not thrown-in a loop break here, it would go on happily counting until the process is manually interrupted or killed (using, for example, ``ctrl-C``).", "_____no_output_____" ], [ "## Useful Iterators\nThis iterator syntax is used nearly universally in Python built-in types as well as the more data science-specific objects we'll explore in later sections.\nHere we'll cover some of the more useful iterators in the Python language:", "_____no_output_____" ], [ "### ``enumerate``\nOften you need to iterate not only the values in an array, but also keep track of the index.\nYou might be tempted to do things this way:", "_____no_output_____" ] ], [ [ "L = [2, 4, 6, 8, 10]\nfor i in range(len(L)):\n print(i, L[i])", "0 2\n1 4\n2 6\n3 8\n4 10\n" ] ], [ [ "Although this does work, Python provides a cleaner syntax using the ``enumerate`` iterator:", "_____no_output_____" ] ], [ [ "for i, val in enumerate(L):\n print(i, val)", "0 2\n1 4\n2 6\n3 8\n4 10\n" ] ], [ [ "This is the more \"Pythonic\" way to enumerate the indices and values in a list.", "_____no_output_____" ], [ "### ``zip``\nOther times, you may have multiple lists that you want to iterate over simultaneously.\nYou could certainly iterate over the index as in the non-Pythonic example we looked at previously, but it is better to use the ``zip`` iterator, which zips together iterables:", "_____no_output_____" ] ], [ [ "L = [2, 4, 6, 8, 10]\nR = [3, 6, 9, 12, 15]\nfor lval, rval in zip(L, R):\n print(lval, rval)", "2 3\n4 6\n6 9\n8 12\n10 15\n" ] ], [ [ "Any number of iterables can be zipped together, and if they are different lengths, the shortest will determine the length of the ``zip``.", "_____no_output_____" ], [ "### ``map`` and ``filter``\nThe ``map`` iterator takes a function and applies it to the values in an iterator:", "_____no_output_____" ] ], [ [ "# find the first 10 square numbers\nsquare = lambda x: x ** 2\nfor val in map(square, range(10)):\n print(val, end=' ')", "0 1 4 9 16 25 36 49 64 81 " ] ], [ [ "The ``filter`` iterator looks similar, except it only passes-through values for which the filter function evaluates to True:", "_____no_output_____" ] ], [ [ "# find values up to 10 for which x % 2 is zero\nis_even = lambda x: x % 2 == 0\nfor val in filter(is_even, range(10)):\n print(val, end=' ')", "0 2 4 6 8 " ] ], [ [ "The ``map`` and ``filter`` functions, along with the ``reduce`` function (which lives in Python's ``functools`` module) are fundamental components of the *functional programming* style, which, while not a dominant programming style in the Python world, has its outspoken proponents (see, for example, the [pytoolz](https://toolz.readthedocs.org/en/latest/) library).", "_____no_output_____" ], [ "### Iterators as function arguments\n\nWe saw in [``*args`` and ``**kwargs``: Flexible Arguments](#*args-and-**kwargs:-Flexible-Arguments). that ``*args`` and ``**kwargs`` can be used to pass sequences and dictionaries to functions.\nIt turns out that the ``*args`` syntax works not just with sequences, but with any iterator:", "_____no_output_____" ] ], [ [ "print(*range(10))", "0 1 2 3 4 5 6 7 8 9\n" ] ], [ [ "So, for example, we can get tricky and compress the ``map`` example from before into the following:", "_____no_output_____" ] ], [ [ "print(*map(lambda x: x ** 2, range(10)))", "0 1 4 9 16 25 36 49 64 81\n" ] ], [ [ "Using this trick lets us answer the age-old question that comes up in Python learners' forums: why is there no ``unzip()`` function which does the opposite of ``zip()``?\nIf you lock yourself in a dark closet and think about it for a while, you might realize that the opposite of ``zip()`` is... ``zip()``! The key is that ``zip()`` can zip-together any number of iterators or sequences. Observe:", "_____no_output_____" ] ], [ [ "L1 = (1, 2, 3, 4)\nL2 = ('a', 'b', 'c', 'd')", "_____no_output_____" ], [ "z = zip(L1, L2)\nprint(*z)", "(1, 'a') (2, 'b') (3, 'c') (4, 'd')\n" ], [ "z = zip(L1, L2)\nnew_L1, new_L2 = zip(*z)\nprint(new_L1, new_L2)", "(1, 2, 3, 4) ('a', 'b', 'c', 'd')\n" ] ], [ [ "Ponder this for a while. If you understand why it works, you'll have come a long way in understanding Python iterators!", "_____no_output_____" ], [ "## Specialized Iterators: ``itertools``\n\nWe briefly looked at the infinite ``range`` iterator, ``itertools.count``.\nThe ``itertools`` module contains a whole host of useful iterators; it's well worth your while to explore the module to see what's available.\nAs an example, consider the ``itertools.permutations`` function, which iterates over all permutations of a sequence:", "_____no_output_____" ] ], [ [ "from itertools import permutations\np = permutations(range(3))\nprint(*p)", "(0, 1, 2) (0, 2, 1) (1, 0, 2) (1, 2, 0) (2, 0, 1) (2, 1, 0)\n" ] ], [ [ "Similarly, the ``itertools.combinations`` function iterates over all unique combinations of ``N`` values within a list:", "_____no_output_____" ] ], [ [ "from itertools import combinations\nc = combinations(range(4), 2)\nprint(*c)", "(0, 1) (0, 2) (0, 3) (1, 2) (1, 3) (2, 3)\n" ] ], [ [ "Somewhat related is the ``product`` iterator, which iterates over all sets of pairs between two or more iterables:", "_____no_output_____" ] ], [ [ "from itertools import product\np = product('ab', range(3))\nprint(*p)", "('a', 0) ('a', 1) ('a', 2) ('b', 0) ('b', 1) ('b', 2)\n" ] ], [ [ "Many more useful iterators exist in ``itertools``: the full list can be found, along with some examples, in Python's [online documentation](https://docs.python.org/3.5/library/itertools.html).", "_____no_output_____" ], [ " ", "_____no_output_____" ], [ "<!--NAVIGATION-->\n< [Errors and Exceptions](09-Errors-and-Exceptions.ipynb) | [Contents](Index.ipynb) | [List Comprehensions](11-List-Comprehensions.ipynb) >", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ] ]

[ "CC-BY-3.0" ]

[ "CC-BY-3.0" ]

[ "CC-BY-3.0" ]

[ [ [ "###### Content under Creative Commons Attribution license CC-BY 4.0, code under MIT license (c)2014 L.A. Barba, G.F. Forsyth, C.D. Cooper.", "_____no_output_____" ], [ "# Spreading out", "_____no_output_____" ], [ "We're back! This is the fourth notebook of _Spreading out: parabolic PDEs,_ Module 4 of the course [**\"Practical Numerical Methods with Python\"**](https://openedx.seas.gwu.edu/courses/course-v1:MAE+MAE6286+2017/about). \n\nIn the [previous notebook](https://nbviewer.jupyter.org/github/numerical-mooc/numerical-mooc/blob/master/lessons/04_spreadout/04_03_Heat_Equation_2D_Explicit.ipynb), we solved a 2D problem for the first time, using an explicit scheme. We know explicit schemes have stability constraints that might make them impractical in some cases, due to requiring a very small time step. Implicit schemes are unconditionally stable, offering the advantage of larger time steps; in [notebook 2](https://nbviewer.jupyter.org/github/numerical-mooc/numerical-mooc/blob/master/lessons/04_spreadout/04_02_Heat_Equation_1D_Implicit.ipynb), we look at the 1D implicit solution of diffusion. Already, that was quite a lot of work: setting up a matrix of coefficients and a right-hand-side vector, while taking care of the boundary conditions, and then solving the linear system. And now, we want to do implicit schemes in 2D—are you ready for this challenge?", "_____no_output_____" ], [ "## 2D Heat conduction", "_____no_output_____" ], [ "We already studied 2D heat conduction in the previous lesson, but now we want to work out how to build an implicit solution scheme. To refresh your memory, here is the heat equation again:\n\n$$\n\\begin{equation}\n\\frac{\\partial T}{\\partial t} = \\alpha \\left(\\frac{\\partial^2 T}{\\partial x^2} + \\frac{\\partial^2 T}{\\partial y^2} \\right)\n\\end{equation}\n$$\n\nOur previous solution used a Dirichlet boundary condition on the left and bottom boundaries, with $T(x=0)=T(y=0)=100$, and a Neumann boundary condition with zero flux on the top and right edges, with $q_x=q_y=0$.\n\n$$\n\\left( \\left.\\frac{\\partial T}{\\partial y}\\right|_{y=0.1} = q_y \\right) \\quad \\text{and} \\quad \\left( \\left.\\frac{\\partial T}{\\partial x}\\right|_{x=0.1} = q_x \\right)\n$$\n\nFigure 1 shows a sketch of the problem set up for our hypothetical computer chip with two hot edges and two insulated edges.", "_____no_output_____" ], [ "#### <img src=\"./figures/2dchip.svg\" width=\"400px\"> Figure 1: Simplified microchip problem setup.", "_____no_output_____" ], [ "### Implicit schemes in 2D", "_____no_output_____" ], [ "An implicit discretization will evaluate the spatial derivatives at the next time level, $t^{n+1}$, using the unknown values of the solution variable. For the 2D heat equation with central difference in space, that is written as:\n\n$$\n\\begin{equation}\n \\begin{split}\n & \\frac{T^{n+1}_{i,j} - T^n_{i,j}}{\\Delta t} = \\\\\n & \\quad \\alpha \\left( \\frac{T^{n+1}_{i+1, j} - 2T^{n+1}_{i,j} + T^{n+1}_{i-1,j}}{\\Delta x^2} + \\frac{T^{n+1}_{i, j+1} - 2T^{n+1}_{i,j} + T^{n+1}_{i,j-1}}{\\Delta y^2} \\right) \\\\\n \\end{split}\n\\end{equation}\n$$\n\nThis equation looks better when we put what we *don't know* on the left and what we *do know* on the right. Make sure to work this out yourself on a piece of paper.\n\n$$\n\\begin{equation}\n \\begin{split}\n & -\\frac{\\alpha \\Delta t}{\\Delta x^2} \\left( T^{n+1}_{i-1,j} + T^{n+1}_{i+1,j} \\right) + \\left( 1 + 2 \\frac{\\alpha \\Delta t}{\\Delta x^2} + 2 \\frac{\\alpha \\Delta t}{\\Delta y^2} \\right) T^{n+1}_{i,j} \\\\\n& \\quad \\quad \\quad -\\frac{\\alpha \\Delta t}{\\Delta y^2} \\left( T^{n+1}_{i,j-1} + T^{n+1}_{i,j+1} \\right) = T^n_{i,j} \\\\\n \\end{split}\n\\end{equation}\n$$\n\nTo make this discussion easier, let's assume that the mesh spacing is the same in both directions and $\\Delta x=\\Delta y = \\delta$:\n\n$$\n\\begin{equation}\n-T^{n+1}_{i-1,j} - T^{n+1}_{i+1,j} + \\left(\\frac{\\delta^2}{\\alpha \\Delta t} + 4 \\right) T^{n+1}_{i,j} - T^{n+1}_{i,j-1}-T^{n+1}_{i,j+1} = \\frac{\\delta^2}{\\alpha \\Delta t}T^n_{i,j}\n\\end{equation}\n$$\n\nJust like in the one-dimensional case, $T_{i,j}$ appears in the equation for $T_{i-1,j}$, $T_{i+1,j}$, $T_{i,j+1}$ and $T_{i,j-1}$, and we can form a linear system to advance in time. But, how do we construct the matrix in this case? What are the $(i+1,j)$, $(i-1,j)$, $(i,j+1)$, and $(i,j-1)$ positions in the matrix?\n\nWith explicit schemes we don't need to worry about these things. We can lay out the data just as it is in the physical problem. We had an array `T` that was a 2-dimensional matrix. To fetch the temperature in the next node in the $x$ direction $(T_{i+1,j})$ we just did `T[j,i+1]`, and likewise in the $y$ direction $(T_{i,j+1})$ was in `T[j+1,i]`. In implicit schemes, we need to think a bit harder about how the data is mapped to the physical problem.\n\nAlso, remember from the [notebook on 1D-implicit schemes](https://nbviewer.jupyter.org/github/numerical-mooc/numerical-mooc/blob/master/lessons/04_spreadout/04_02_Heat_Equation_1D_Implicit.ipynb) that the linear system had $N-2$ elements? We applied boundary conditions on nodes $i=0$ and $i=N-1$, and they were not modified by the linear system. In 2D, this becomes a bit more complicated.\n\nLet's use Figure 1, representing a set of grid nodes in two dimensions, to guide the discussion.", "_____no_output_____" ], [ "#### <img src=\"./figures/2D_discretization.png\"> Figure 2: Layout of matrix elements in 2D problem", "_____no_output_____" ], [ "Say we have the 2D domain of size $L_x\\times L_y$ discretized in $n_x$ and $n_y$ points. We can divide the nodes into boundary nodes (empty circles) and interior nodes (filled circles).\n\nThe boundary nodes, as the name says, are on the boundary. They are the nodes with indices $(i=0,j)$, $(i=n_x-1,j)$, $(i,j=0)$, and $(i,j=n_y-1)$, and boundary conditions are enforced there.\n\nThe interior nodes are not on the boundary, and the finite-difference equation acts on them. If we leave the boundary nodes aside for the moment, then the grid will have $(n_x-2)\\cdot(n_y-2)$ nodes that need to be updated on each time step. This is the number of unknowns in the linear system. The matrix of coefficients will have $\\left( (n_x-2)\\cdot(n_y-2) \\right)^2$ elements (most of them zero!).\n\nTo construct the matrix, we will iterate over the nodes in an x-major order: index $i$ will run faster. The order will be \n\n* $(i=1,j=1)$\n* $(i=2,j=1)$ ...\n* $(i=nx-2,j=1)$\n* $(i=1,j=2)$\n* $(i=2,j=2)$ ... \n* $(i=n_x-2,j=n_y-2)$. \n\nThat is the ordering represented by dotted line on Figure 1. Of course, if you prefer to organize the nodes differently, feel free to do so!\n\nBecause we chose this ordering, the equation for nodes $(i-1,j)$ and $(i+1,j)$ will be just before and after $(i,j)$, respectively. But what about $(i,j-1)$ and $(i,j+1)$? Even though in the physical problem they are very close, the equations are $n_x-2$ places apart! This can tie your head in knots pretty quickly. \n\n_The only way to truly understand it is to make your own diagrams and annotations on a piece of paper and reconstruct this argument!_", "_____no_output_____" ], [ "### Boundary conditions", "_____no_output_____" ], [ "Before we attempt to build the matrix, we need to think about boundary conditions. There is some bookkeeping to be done here, so bear with us for a moment.\n\nSay, for example, that the left and bottom boundaries have Dirichlet boundary conditions, and the top and right boundaries have Neumann boundary conditions.\n\nLet's look at each case:\n\n**Bottom boundary:**\n \nThe equation for $j=1$ (interior points adjacent to the bottom boundary) uses values from $j=0$, which are known. Let's put that on the right-hand side of the equation. We get this equation for all points across the $x$-axis that are adjacent to the bottom boundary:\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{i-1,1} - T^{n+1}_{i+1,1} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 4 \\right) T^{n+1}_{i,1} - T^{n+1}_{i,j+1} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{i,1} + T^{n+1}_{i,0} & \\\\\n \\end{split}\n\\end{equation}\n$$\n\n**Left boundary:**\n\nLike for the bottom boundary, the equation for $i=1$ (interior points adjacent to the left boundary) uses known values from $i=0$, and we will put that on the right-hand side:\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{2,j} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 4 \\right) T^{n+1}_{1,j} - T^{n+1}_{1,j-1} - T^{n+1}_{1,j+1} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{1,j} + T^{n+1}_{0,j} & \\\\\n \\end{split}\n\\end{equation}\n$$\n\n**Right boundary:**\n\nSay the boundary condition is $\\left. \\frac{\\partial T}{\\partial x} \\right|_{x=L_x} = q_x$. Its finite-difference approximation is\n\n$$\n\\begin{equation}\n \\frac{T^{n+1}_{n_x-1,j} - T^{n+1}_{n_x-2,j}}{\\delta} = q_x\n\\end{equation}\n$$\n\nWe can write $T^{n+1}_{n_x-1,j} = \\delta q_x + T^{n+1}_{n_x-2,j}$ to get the finite difference equation for $i=n_x-2$:\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{n_x-3,j} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 3 \\right) T^{n+1}_{n_x-2,j} - T^{n+1}_{n_x-2,j-1} - T^{n+1}_{n_x-2,j+1} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{n_x-2,j} + \\delta q_x & \\\\\n \\end{split}\n\\end{equation}\n$$\n\nNot sure about this? Grab pen and paper! _Please_, check this yourself. It will help you understand!\n\n**Top boundary:**\n\nNeumann boundary conditions specify the derivative normal to the boundary: $\\left. \\frac{\\partial T}{\\partial y} \\right|_{y=L_y} = q_y$. No need to repeat what we did for the right boundary, right? The equation for $j=n_y-2$ is\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{i-1,n_y-2} - T^{n+1}_{i+1,n_y-2} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 3 \\right) T^{n+1}_{i,n_y-2} - T^{n+1}_{i,n_y-3} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{i,n_y-2} + \\delta q_y & \\\\\n \\end{split}\n\\end{equation}\n$$\n\nSo far, we have then 5 possible cases: bottom, left, right, top, and interior points. Does this cover everything? What about corners?", "_____no_output_____" ], [ "**Bottom-left corner**\n\nAt $T_{1,1}$ there is a Dirichlet boundary condition at $i=0$ and $j=0$. This equation is:\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{2,1} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 4 \\right) T^{n+1}_{1,1} - T^{n+1}_{1,2} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{1,1} + T^{n+1}_{0,1} + T^{n+1}_{1,0} & \\\\\n \\end{split}\n\\end{equation}\n$$\n\n**Top-left corner:**\n\nAt $T_{1,n_y-2}$ there is a Dirichlet boundary condition at $i=0$ and a Neumann boundary condition at $i=n_y-1$. This equation is:\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{2,n_y-2} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 3 \\right) T^{n+1}_{1,n_y-2} - T^{n+1}_{1,n_y-3} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{1,n_y-2} + T^{n+1}_{0,n_y-2} + \\delta q_y & \\\\\n \\end{split}\n\\end{equation}\n$$\n\n**Top-right corner**\n\nAt $T_{n_x-2,n_y-2}$, there are Neumann boundary conditions at both $i=n_x-1$ and $j=n_y-1$. The finite difference equation is then\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{n_x-3,n_y-2} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 2 \\right) T^{n+1}_{n_x-2,n_y-2} - T^{n+1}_{n_x-2,n_y-3} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{n_x-2,n_y-2} + \\delta(q_x + q_y) & \\\\\n \\end{split}\n\\end{equation}\n$$\n\n**Bottom-right corner**\n\nTo calculate $T_{n_x-2,1}$ we need to consider a Dirichlet boundary condition to the bottom and a Neumann boundary condition to the right. We will get a similar equation to the top-left corner!\n\n$$\n\\begin{equation}\n \\begin{split}\n -T^{n+1}_{n_x-3,1} + \\left( \\frac{\\delta^2}{\\alpha \\Delta t} + 3 \\right) T^{n+1}_{n_x-2,1} - T^{n+1}_{n_x-2,2} \\qquad & \\\\\n = \\frac{\\delta^2}{\\alpha \\Delta t} T^n_{n_x-2,1} + T^{n+1}_{n_x-2,0} + \\delta q_x & \\\\\n \\end{split}\n\\end{equation}\n$$\n\nOkay, now we are actually ready. We have checked every possible case!", "_____no_output_____" ], [ "### The linear system", "_____no_output_____" ], [ "Like in the previous lesson introducing implicit schemes, we will solve a linear system at every time step:\n\n$$\n[A][T^{n+1}_\\text{int}] = [b]+[b]_{b.c.}\n$$\n\nThe coefficient matrix now takes some more work to figure out and to build in code. There is no substitute for you working this out patiently on paper!\n\nThe structure of the matrix can be described as a series of diagonal blocks, and lots of zeros elsewhere. Look at Figure 3, representing the block structure of the coefficient matrix, and refer back to Figure 2, showing the discretization grid in physical space. The first row of interior points, adjacent to the bottom boundary, generates the matrix block labeled $A_1$. The top row of interior points, adjacent to the top boundary generates the matrix block labeled $A_3$. All other interior points in the grid generate similar blocks, labeled $A_2$ on Figure 3.", "_____no_output_____" ], [ "#### <img src=\"./figures/implicit-matrix-blocks.png\"> Figure 3: Sketch of coefficient-matrix blocks.", "_____no_output_____" ], [ "#### <img src=\"./figures/matrix-blocks-on-grid.png\"> Figure 4: Grid points corresponding to each matrix-block type.", "_____no_output_____" ], [ "The matrix block $A_1$ is\n\n<img src=\"./figures/A_1.svg\" width=\"640px\">\n\nThe block matrix $A_2$ is\n\n<img src=\"./figures/A_2.svg\" width=\"640px\">\n\nThe block matrix $A_3$ is\n\n<img src=\"./figures/A_3.svg\" width=\"640px\">", "_____no_output_____" ], [ "Vector $T^{n+1}_\\text{int}$ contains the temperature of the interior nodes in the next time step. It is:\n\n$$\n\\begin{equation}\nT^{n+1}_\\text{int} = \\left[\n\\begin{array}{c}\nT^{n+1}_{1,1}\\\\\nT^{n+1}_{2,1} \\\\\n\\vdots \\\\\nT^{n+1}_{n_x-2,1} \\\\\nT^{n+1}_{2,1} \\\\\n\\vdots \\\\\nT^{n+1}_{n_x-2,n_y-2}\n\\end{array}\n\\right]\n\\end{equation}\n$$\n\nRemember the x-major ordering we chose!", "_____no_output_____" ], [ "Finally, the right-hand side is\n\\begin{equation}\n[b]+[b]_{b.c.} = \n\\left[\\begin{array}{c}\n\\sigma^\\prime T^n_{1,1} + T^{n+1}_{0,1} + T^{n+1}_{1,0} \\\\\n\\sigma^\\prime T^n_{2,0} + T^{n+1}_{2,0} \\\\\n\\vdots \\\\\n\\sigma^\\prime T^n_{n_x-2,1} + T^{n+1}_{n_x-2,0} + \\delta q_x \\\\\n\\sigma^\\prime T^n_{1,2} + T^{n+1}_{0,2} \\\\\n\\vdots \\\\\n\\sigma^\\prime T^n_{n_x-2,n_y-2} + \\delta(q_x + q_y)\n\\end{array}\\right]\n\\end{equation}", "_____no_output_____" ], [ "where $\\sigma^\\prime = 1/\\sigma = \\delta^2/\\alpha \\Delta t$. The matrix looks very ugly, but it is important you understand it! Think about it. Can you answer:\n * Why a -1 factor appears $n_x-2$ columns after the diagonal? What about $n_x-2$ columns before the diagonal?\n * Why in row $n_x-2$ the position after the diagonal contains a 0?\n * Why in row $n_x-2$ the diagonal is $\\sigma^\\prime + 3$ rather than $\\sigma^\\prime + 4$?\n * Why in the last row the diagonal is $\\sigma^\\prime + 2$ rather than $\\sigma^\\prime + 4$?\n \nIf you can answer those questions, you are in good shape to continue!", "_____no_output_____" ], [ "Let's write a function that will generate the matrix and right-hand side for the heat conduction problem in the previous notebook. Remember, we had Dirichlet boundary conditions in the left and bottom, and zero-flux Neumann boundary condition on the top and right $(q_x=q_y=0)$. \n\nAlso, we'll import `scipy.linalg.solve` because we need to solve a linear system.", "_____no_output_____" ] ], [ [ "import numpy\nfrom scipy import linalg", "_____no_output_____" ], [ "def lhs_operator(M, N, sigma):\n \"\"\"\n Assembles and returns the implicit operator\n of the system for the 2D diffusion equation.\n We use a Dirichlet condition at the left and\n bottom boundaries and a Neumann condition\n (zero-gradient) at the right and top boundaries.\n \n Parameters\n ----------\n M : integer\n Number of interior points in the x direction.\n N : integer\n Number of interior points in the y direction.\n sigma : float\n Value of alpha * dt / dx**2.\n \n Returns\n -------\n A : numpy.ndarray\n The implicit operator as a 2D array of floats\n of size M*N by M*N.\n \"\"\"\n A = numpy.zeros((M * N, M * N))\n for j in range(N):\n for i in range(M):\n I = j * M + i # row index\n # Get index of south, west, east, and north points.\n south, west, east, north = I - M, I - 1, I + 1, I + M\n # Setup coefficients at corner points.\n if i == 0 and j == 0: # bottom-left corner\n A[I, I] = 1.0 / sigma + 4.0\n A[I, east] = -1.0\n A[I, north] = -1.0\n elif i == M - 1 and j == 0: # bottom-right corner\n A[I, I] = 1.0 / sigma + 3.0\n A[I, west] = -1.0\n A[I, north] = -1.0\n elif i == 0 and j == N - 1: # top-left corner\n A[I, I] = 1.0 / sigma + 3.0\n A[I, south] = -1.0\n A[I, east] = -1.0\n elif i == M - 1 and j == N - 1: # top-right corner\n A[I, I] = 1.0 / sigma + 2.0\n A[I, south] = -1.0\n A[I, west] = -1.0\n # Setup coefficients at side points (excluding corners).\n elif i == 0: # left side\n A[I, I] = 1.0 / sigma + 4.0\n A[I, south] = -1.0\n A[I, east] = -1.0\n A[I, north] = -1.0\n elif i == M - 1: # right side\n A[I, I] = 1.0 / sigma + 3.0\n A[I, south] = -1.0\n A[I, west] = -1.0\n A[I, north] = -1.0\n elif j == 0: # bottom side\n A[I, I] = 1.0 / sigma + 4.0\n A[I, west] = -1.0\n A[I, east] = -1.0\n A[I, north] = -1.0\n elif j == N - 1: # top side\n A[I, I] = 1.0 / sigma + 3.0\n A[I, south] = -1.0\n A[I, west] = -1.0\n A[I, east] = -1.0\n # Setup coefficients at interior points.\n else:\n A[I, I] = 1.0 / sigma + 4.0\n A[I, south] = -1.0\n A[I, west] = -1.0\n A[I, east] = -1.0\n A[I, north] = -1.0\n return A", "_____no_output_____" ], [ "def rhs_vector(T, M, N, sigma, Tb):\n \"\"\"\n Assembles and returns the right-hand side vector\n of the system for the 2D diffusion equation.\n We use a Dirichlet condition at the left and\n bottom boundaries and a Neumann condition\n (zero-gradient) at the right and top boundaries.\n \n Parameters\n ----------\n T : numpy.ndarray\n The temperature distribution as a 1D array of floats.\n M : integer\n Number of interior points in the x direction.\n N : integer\n Number of interior points in the y direction.\n sigma : float\n Value of alpha * dt / dx**2.\n Tb : float\n Boundary value for Dirichlet conditions.\n \n Returns\n -------\n b : numpy.ndarray\n The right-hand side vector as a 1D array of floats\n of size M*N.\n \"\"\"\n b = 1.0 / sigma * T\n # Add Dirichlet term at points located next\n # to the left and bottom boundaries.\n for j in range(N):\n for i in range(M):\n I = j * M + i\n if i == 0:\n b[I] += Tb\n if j == 0:\n b[I] += Tb\n return b", "_____no_output_____" ] ], [ [ "The solution of the linear system $(T^{n+1}_\\text{int})$ contains the temperatures of the interior points at the next time step in a 1D array. We will also create a function that will take the values of $T^{n+1}_\\text{int}$ and put them in a 2D array that resembles the physical domain.", "_____no_output_____" ] ], [ [ "def map_1d_to_2d(T_1d, nx, ny, Tb):\n \"\"\"\n Maps a 1D array of the temperature at the interior points\n to a 2D array that includes the boundary values.\n \n Parameters\n ----------\n T_1d : numpy.ndarray\n The temperature at the interior points as a 1D array of floats.\n nx : integer\n Number of points in the x direction of the domain.\n ny : integer\n Number of points in the y direction of the domain.\n Tb : float\n Boundary value for Dirichlet conditions.\n \n Returns\n -------\n T : numpy.ndarray\n The temperature distribution in the domain\n as a 2D array of size ny by nx.\n \"\"\"\n T = numpy.zeros((ny, nx))\n # Get the value at interior points.\n T[1:-1, 1:-1] = T_1d.reshape((ny - 2, nx - 2))\n # Use Dirichlet condition at left and bottom boundaries.\n T[:, 0] = Tb\n T[0, :] = Tb\n # Use Neumann condition at right and top boundaries.\n T[:, -1] = T[:, -2]\n T[-1, :] = T[-2, :]\n return T", "_____no_output_____" ] ], [ [ "And to advance in time, we will use", "_____no_output_____" ] ], [ [ "def btcs_implicit_2d(T0, nt, dt, dx, alpha, Tb):\n \"\"\"\n Computes and returns the distribution of the\n temperature after a given number of time steps.\n \n The 2D diffusion equation is integrated using\n Euler implicit in time and central differencing\n in space, with a Dirichlet condition at the left\n and bottom boundaries and a Neumann condition\n (zero-gradient) at the right and top boundaries.\n \n Parameters\n ----------\n T0 : numpy.ndarray\n The initial temperature distribution as a 2D array of floats.\n nt : integer\n Number of time steps to compute.\n dt : float\n Time-step size.\n dx : float\n Grid spacing in the x and y directions.\n alpha : float\n Thermal diffusivity of the plate.\n Tb : float\n Boundary value for Dirichlet conditions.\n \n Returns\n -------\n T : numpy.ndarray\n The temperature distribution as a 2D array of floats.\n \"\"\"\n # Get the number of points in each direction.\n ny, nx = T0.shape\n # Get the number of interior points in each direction.\n M, N = nx - 2, ny - 2\n # Compute the constant sigma.\n sigma = alpha * dt / dx**2\n # Create the implicit operator of the system.\n A = lhs_operator(M, N, sigma)\n # Integrate in time.\n T = T0[1:-1, 1:-1].flatten() # interior points as a 1D array\n I, J = int(M / 2), int(N / 2) # indices of the center\n for n in range(nt):\n # Compute the right-hand side of the system.\n b = rhs_vector(T, M, N, sigma, Tb)\n # Solve the system with scipy.linalg.solve.\n T = linalg.solve(A, b)\n # Check if the center of the domain has reached T = 70C.\n if T[J * M + I] >= 70.0:\n break\n print('[time step {}] Center at T={:.2f} at t={:.2f} s'\n .format(n + 1, T[J * M + I], (n + 1) * dt))\n # Returns the temperature in the domain as a 2D array.\n return map_1d_to_2d(T, nx, ny, Tb)", "_____no_output_____" ] ], [ [ "Remember, we want the function to tell us when the center of the plate reaches $70^\\circ C$.", "_____no_output_____" ], [ "##### Dig deeper", "_____no_output_____" ], [ "For demonstration purposes, these functions are very explicit. But you can see a trend here, right? \n\nSay we start with a matrix with `1/sigma+4` in the main diagonal, and `-1` on the 4 other corresponding diagonals. Now, we have to modify the matrix only where the boundary conditions are affecting. We saw the impact of the Dirichlet and Neumann boundary condition on each position of the matrix, we just need to know in which position to perform those changes. \n\nA function that maps `i` and `j` into `row_number` would be handy, right? How about `row_number = (j-1)*(nx-2)+(i-1)`? By feeding `i` and `j` to that equation, you know exactly where to operate on the matrix. For example, `i=nx-2, j=2`, which is in row `row_number = 2*nx-5`, is next to a Neumann boundary condition: we have to substract one out of the main diagonal (`A[2*nx-5,2*nx-5]-=1`), and put a zero in the next column (`A[2*nx-5,2*nx-4]=0`). This way, the function can become much simpler!\n\nCan you use this information to construct a more general function `lhs_operator`? Can you make it such that the type of boundary condition is an input to the function? ", "_____no_output_____" ], [ "## Heat diffusion in 2D", "_____no_output_____" ], [ "Let's recast the 2D heat conduction from the previous notebook, and solve it with an implicit scheme. ", "_____no_output_____" ] ], [ [ "# Set parameters.\nLx = 0.01 # length of the plate in the x direction\nLy = 0.01 # length of the plate in the y direction\nnx = 21 # number of points in the x direction\nny = 21 # number of points in the y direction\ndx = Lx / (nx - 1) # grid spacing in the x direction\ndy = Ly / (ny - 1) # grid spacing in the y direction\nalpha = 1e-4 # thermal diffusivity\n\n# Define the locations along a gridline.\nx = numpy.linspace(0.0, Lx, num=nx)\ny = numpy.linspace(0.0, Ly, num=ny)\n\n# Compute the initial temperature distribution.\nTb = 100.0 # temperature at the left and bottom boundaries\nT0 = 20.0 * numpy.ones((ny, nx))\nT0[:, 0] = Tb\nT0[0, :] = Tb", "_____no_output_____" ] ], [ [ "We are ready to go!", "_____no_output_____" ] ], [ [ "# Set the time-step size based on CFL limit.\nsigma = 0.25\ndt = sigma * min(dx, dy)**2 / alpha # time-step size\nnt = 300 # number of time steps to compute\n\n# Compute the temperature along the rod.\nT = btcs_implicit_2d(T0, nt, dt, dx, alpha, Tb)", "[time step 257] Center at T=70.00 at t=0.16 s\n" ] ], [ [ "And plot,", "_____no_output_____" ] ], [ [ "from matplotlib import pyplot\n%matplotlib inline", "_____no_output_____" ], [ "# Set the font family and size to use for Matplotlib figures.\npyplot.rcParams['font.family'] = 'serif'\npyplot.rcParams['font.size'] = 16", "_____no_output_____" ], [ "# Plot the filled contour of the temperature.\npyplot.figure(figsize=(8.0, 5.0))\npyplot.xlabel('x [m]')\npyplot.ylabel('y [m]')\nlevels = numpy.linspace(20.0, 100.0, num=51)\ncontf = pyplot.contourf(x, y, T, levels=levels)\ncbar = pyplot.colorbar(contf)\ncbar.set_label('Temperature [C]')\npyplot.axis('scaled', adjustable='box');", "_____no_output_____" ] ], [ [ "Try this out with different values of `sigma`! You'll see that it will always give a stable solution!\n\nDoes this result match the explicit scheme from the previous notebook? Do they take the same amount of time to reach $70^\\circ C$ in the center of the plate? Now that we can use higher values of `sigma`, we need fewer time steps for the center of the plate to reach $70^\\circ C$! Of course, we need to be careful that `dt` is small enough to resolve the physics correctly.", "_____no_output_____" ], [ "---\n###### The cell below loads the style of the notebook", "_____no_output_____" ] ], [ [ "from IPython.core.display import HTML\ncss_file = '../../styles/numericalmoocstyle.css'\nHTML(open(css_file, 'r').read())", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import re", "_____no_output_____" ], [ "s = '[email protected], [email protected], [email protected]'", "_____no_output_____" ], [ "result = re.sub(r'[a-z]+@[a-z]+\\.com', 'new-address', s)\nprint(result)", "new-address, new-address, [email protected]\n" ], [ "print(type(result))", "<class 'str'>\n" ], [ "result = re.sub(r'([a-z]+)@([a-z]+)\\.com', r'\\1@\\2.net', s)\nprint(result)", "[email protected], [email protected], [email protected]\n" ], [ "result = re.sub(r'(?P<local>[a-z]+)@(?P<SLD>[a-z]+)\\.com', r'\\g<local>@\\g<SLD>.net', s)\nprint(result)", "[email protected], [email protected], [email protected]\n" ], [ "result = re.sub(r'[a-z]+@[a-z]+\\.com', 'new-address', s, count=1)\nprint(result)", "new-address, [email protected], [email protected]\n" ], [ "result = re.subn(r'[a-z]+@[a-z]+\\.com', 'new-address', s)\nprint(result)", "('new-address, new-address, [email protected]', 2)\n" ], [ "print(result[0])", "new-address, new-address, [email protected]\n" ], [ "print(result[1])", "2\n" ], [ "result = re.subn(r'(?P<local>[a-z]+)@(?P<SLD>[a-z]+)\\.com', r'\\g<local>@\\g<SLD>.net', s)\nprint(result)", "('[email protected], [email protected], [email protected]', 2)\n" ], [ "result = re.subn(r'[a-z]+@[a-z]+\\.com', 'new-address', s, count=1)\nprint(result)", "('new-address, [email protected], [email protected]', 1)\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# DATA 512: A1 Data Curation Assignment\nBy: Megan Nalani Chun", "_____no_output_____" ], [ "## Step 1: Gathering the data <br>\nGather Wikipedia traffic from Jan 1, 2008 - August 30, 2020 <br>\n- Legacy Pagecounts API provides desktop and mobile traffic data from Dec. 2007 - July 2016 <br>\n- Pageviews API provides desktop, mobile web, and mobile app traffic data from July 2015 - last month. \n", "_____no_output_____" ], [ "First, import the json and requests libraries to call the Pagecounts and Pageviews APIs and save the output in json format.", "_____no_output_____" ] ], [ [ "import json\nimport requests", "_____no_output_____" ] ], [ [ "Second, set the location of the endpoints and header information. This information is needed to call the Pagecounts and Pageviews APIs.", "_____no_output_____" ] ], [ [ "endpoint_legacy = 'https://wikimedia.org/api/rest_v1/metrics/legacy/pagecounts/aggregate/{project}/{access-site}/{granularity}/{start}/{end}'\n\nendpoint_pageviews = 'https://wikimedia.org/api/rest_v1/metrics/pageviews/aggregate/{project}/{access}/{agent}/{granularity}/{start}/{end}'\n\nheaders = {\n 'User-Agent': 'https://github.com/NalaniKai/',\n 'From': '[email protected]'\n}", "_____no_output_____" ] ], [ [ "Third, define a function to call the APIs taking in the endpoint and parameters. This function returns the data in json format. ", "_____no_output_____" ] ], [ [ "def api_call(endpoint, parameters):\n call = requests.get(endpoint.format(**parameters), headers=headers)\n response = call.json()\n \n return response", "_____no_output_____" ] ], [ [ "Fourth, define a function to make the api_call() with the correct parameters and save the output to json file in format apiname_accesstype_startyearmonth_endyearmonth.json", "_____no_output_____" ] ], [ [ "def get_data(api, access_dict, params, endpoint, access_name):\n start = 0 #index of start date\n end = 1 #index of end data\n year_month = 6 #size of YYYYMM\n for access_type, start_end_dates in access_dict.items(): #get data for all access types in API\n params[access_name] = access_type\n data = api_call(endpoint, params)\n with open(api + \"_\" + access_type + \"_\" + start_end_dates[start][:year_month] + \"_\" + start_end_dates[end][:year_month] + \".json\", 'w') as f:\n json.dump(data, f) #save data", "_____no_output_____" ] ], [ [ "Fifth, define the parameters for the legacy page count API and call the get_data function. This will pull the data and save it.", "_____no_output_____" ] ], [ [ "api = \"pagecounts\"\naccess_type_legacy = {\"desktop-site\": [\"2008010100\", \"2016070100\"], #access type: start year_month, end year_month\n \"mobile-site\": [\"2014100100\", \"2016070100\"]} #used to save outputs with correct filenames\n#https://wikimedia.org/api/rest_v1/#!/Legacy_data/get_metrics_legacy_pagecounts_aggregate_project_access_site_granularity_start_end\nparams_legacy = {\"project\" : \"en.wikipedia.org\",\n \"granularity\" : \"monthly\",\n \"start\" : \"2008010100\",\n \"end\" : \"2016080100\" #will get data through July 2016\n }\nget_data(api, access_type_legacy, params_legacy, endpoint_legacy, \"access-site\")", "_____no_output_____" ] ], [ [ "Sixth, define the parameters for the page views API and call the get_data function. This will pull the data and save it.", "_____no_output_____" ] ], [ [ "api = \"pageviews\"\nstart_end_dates = [\"2015070100\", \"2020080100\"] #start year_month, end year_month\naccess_type_pageviews = {\"desktop\": start_end_dates, #access type: start year_month, end year_month\n \"mobile-app\": start_end_dates, \n \"mobile-web\": start_end_dates }\n#https://wikimedia.org/api/rest_v1/#!/Pageviews_data/get_metrics_pageviews_aggregate_project_access_agent_granularity_start_end\nparams_pageviews = {\"project\" : \"en.wikipedia.org\",\n \"access\" : \"mobile-web\",\n \"agent\" : \"user\", #remove crawler traffic\n \"granularity\" : \"monthly\",\n \"start\" : \"2008010100\",\n \"end\" : '2020090100' #will get data through August 2020\n }\nget_data(api, access_type_pageviews, params_pageviews, endpoint_pageviews, \"access\") ", "_____no_output_____" ] ], [ [ "## Step 2: Processing the data", "_____no_output_____" ], [ "First, create a function to read in all data files and extract the list of records from items.", "_____no_output_____" ] ], [ [ "def read_json(filename):\n with open(filename, 'r') as f:\n return json.load(f)[\"items\"]", "_____no_output_____" ] ], [ [ "Second, use the read_json function to get a list of records for each file.", "_____no_output_____" ] ], [ [ "pageviews_mobile_web = read_json(\"pageviews_mobile-web_201507_202008.json\")\npageviews_mobile_app = read_json(\"pageviews_mobile-app_201507_202008.json\")\npageviews_desktop = read_json(\"pageviews_desktop_201507_202008.json\")\npagecounts_mobile = read_json(\"pagecounts_mobile-site_201410_201607.json\")\npagecounts_desktop = read_json(\"pagecounts_desktop-site_200801_201607.json\")", "_____no_output_____" ] ], [ [ "Third, create a total mobile traffic count for each month using the mobile-app and mobile-web data from the pageviews API. The list of [timestamp, view_count] pairs data structure will enable easy transformation into a dataframe.", "_____no_output_____" ] ], [ [ "pageviews_mobile = [[r1[\"timestamp\"], r0[\"views\"] + r1[\"views\"]] for r0 in pageviews_mobile_web for r1 in pageviews_mobile_app if r0[\"timestamp\"] == r1[\"timestamp\"]]", "_____no_output_____" ] ], [ [ "Fourth, get the timestamps and values in the [timestamp, view_count] format for the desktop pageviews, desktop pagecounts, and mobile pagecounts. ", "_____no_output_____" ] ], [ [ "pageviews_desktop = [[record[\"timestamp\"], record[\"views\"]] for record in pageviews_desktop]\npagecounts_desktop = [[record[\"timestamp\"], record[\"count\"]] for record in pagecounts_desktop]\npagecounts_mobile = [[record[\"timestamp\"], record[\"count\"]] for record in pagecounts_mobile]", "_____no_output_____" ] ], [ [ "Fifth, import pandas library and transform data into dataframes.", "_____no_output_____" ] ], [ [ "import pandas as pd \npageview_desktop_views = pd.DataFrame(pageviews_desktop, columns=[\"timestamp\", \"pageview_desktop_views\"])\npageview_mobile_views = pd.DataFrame(pageviews_mobile, columns=[\"timestamp\", \"pageview_mobile_views\"])\npagecounts_desktop = pd.DataFrame(pagecounts_desktop, columns=[\"timestamp\", \"pagecount_desktop_views\"])\npagecounts_mobile = pd.DataFrame(pagecounts_mobile, columns=[\"timestamp\", \"pagecount_mobile_views\"])", "_____no_output_____" ] ], [ [ "Sixth, join page view dataframes and calculate total for all views.", "_____no_output_____" ] ], [ [ "df_pageviews = pd.merge(pageview_desktop_views, pageview_mobile_views, how=\"outer\", on=\"timestamp\")\ndf_pageviews[\"pageview_all_views\"] = df_pageviews[\"pageview_desktop_views\"] + df_pageviews[\"pageview_mobile_views\"]\ndf_pageviews.head()", "_____no_output_____" ] ], [ [ "Seventh, join page count dataframes. Then fill in NaN values with 0 to calculate total for all counts.", "_____no_output_____" ] ], [ [ "df_pagecounts = pd.merge(pagecounts_desktop, pagecounts_mobile, how=\"outer\", on=\"timestamp\")\ndf_pagecounts[\"pagecount_mobile_views\"] = df_pagecounts[\"pagecount_mobile_views\"].fillna(0)\ndf_pagecounts[\"pagecount_all_views\"] = df_pagecounts[\"pagecount_desktop_views\"] + df_pagecounts[\"pagecount_mobile_views\"]\ndf_pagecounts.head()", "_____no_output_____" ] ], [ [ "Eighth, join page count and page view dataframes into one table. Filling in missing values with 0.", "_____no_output_____" ] ], [ [ "df = pd.merge(df_pagecounts, df_pageviews, how=\"outer\", on=\"timestamp\")\ndf = df.fillna(0)\ndf.head()", "_____no_output_____" ] ], [ [ "Ninth, separate the timestamp into the year and month in YYYY and MM format for all the data. Remove the timestamp column.", "_____no_output_____" ] ], [ [ "df[\"year\"] = df[\"timestamp\"].apply(lambda date: date[:4])\ndf[\"month\"] = df[\"timestamp\"].apply(lambda date: date[4:6])\ndf.drop(\"timestamp\", axis=1, inplace=True)\ndf.head()", "_____no_output_____" ] ], [ [ "Tenth, save processed data to csv file without the index column. ", "_____no_output_____" ] ], [ [ "df.to_csv(\"en-wikipedia_traffic_200801-202008.csv\", index=False)", "_____no_output_____" ] ], [ [ "## Step 3: Analyze the data", "_____no_output_____" ], [ "First, fill 0 values with numpy.nan values so these values are not plotted on the chart.", "_____no_output_____" ] ], [ [ "import numpy as np\ndf.replace(0, np.nan, inplace=True)", "_____no_output_____" ] ], [ [ "Second, transform the year and month into a datetime.date type which will be used for the x-axis in the chart.", "_____no_output_____" ] ], [ [ "from datetime import date\ndate = df.apply(lambda r: date(int(r[\"year\"]), int(r[\"month\"]), 1), axis=1) ", "_____no_output_____" ] ], [ [ "Third, divide all page view counts by 1e6 so the chart is easier to read. Y-axis will be the values shown x 1,000,000.", "_____no_output_____" ] ], [ [ "pc_mobile = df[\"pagecount_mobile_views\"] / 1e6\npv_mobile = df[\"pageview_mobile_views\"] / 1e6\npc_desktop = df[\"pagecount_desktop_views\"] / 1e6\npv_desktop = df[\"pageview_desktop_views\"] / 1e6\npv_total = df[\"pageview_all_views\"] / 1e6\npc_total = df[\"pagecount_all_views\"] / 1e6", "_____no_output_____" ] ], [ [ "Fourth, plot the data in a time series for desktop (main site), mobile, and the total all up. The dashed lines are data from the pagecount API and the solid lines are the data from the pageview API without crawler traffic.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt \nfrom matplotlib.dates import YearLocator\n\n#create plot and assign time series to plot\nfig, ax1 = plt.subplots(figsize=(18,6))\nax1.plot(date, pc_desktop, label=\"main site\", color=\"green\", ls=\"--\")\nax1.plot(date, pv_desktop, label=\"_Hidden label\", color=\"green\", ls=\"-\")\nax1.plot(date, pc_mobile, label=\"mobile site\", color=\"blue\", ls=\"--\")\nax1.plot(date, pv_mobile, label=\"_Hidden label\", color=\"blue\", ls=\"-\")\nax1.plot(date, pc_total, label=\"total\", color=\"black\", ls=\"--\")\nax1.plot(date, pv_total, label=\"_Hidden label\", color=\"black\", ls=\"-\")\n\nax1.xaxis.set_major_locator(YearLocator()) #show every year on the x-axis\n\n#set caption\ncaption = \"May 2015: a new pageview definition took effect, which eliminated all crawler traffic. Solid lines mark new definition.\"\nfig.text(.5, .01, caption, ha='center', color=\"red\")\n\n#set labels for x-axis, y-axis, and title\nplt.xlabel(\"Date\")\nplt.ylabel(\"Page Views (x 1,000,000)\")\nplt.title(\"Page Views on English Wikipedia (x 1,000,000)\")\n\nplt.ylim(ymin=0) #start y-axis at 0\n\nplt.grid(True) #turn on background grid\nplt.legend(loc=\"upper left\")\n\n#save chart to png file\nfilename = \"Time Series.png\"\nplt.savefig(filename)\n\nplt.show() #display chart", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "## İki saniye periyotlu kare dalga\n### Sekil 3.24\n", "_____no_output_____" ] ], [ [ "# Tanimlar\nimport matplotlib.pyplot as plt\nimport numpy as np\n\nt=np.linspace(0,10, 1000)\nfreqs=np.r_[0.5:6.5]\n\nfig, ax=plt.subplots(7,1, sharex=True, sharey=True)\nsumSines=np.zeros(t.shape);\ncnt=1;\nfor f in freqs:\n s= np.sin(2*np.pi*f*t)/(2*f)\n ax[cnt].plot(t,s)\n sumSines+=s\n cnt+=1\n \nax[0].plot(t,sumSines, 'r')\nplt.show()", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import tensorflow\nimport pickle\nfrom tensorflow.keras.utils import to_categorical\nimport numpy as np\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.utils import shuffle\nfrom src.dataloader import IMDB\nfrom sklearn.model_selection import StratifiedShuffleSplit\nfrom sklearn.metrics import confusion_matrix\nfrom src.model import get_model\nfrom src.embedding import *", "_____no_output_____" ] ], [ [ "# load data and embedding ", "_____no_output_____" ] ], [ [ "# load confing\nimport yaml\nimport pickle\nwith open('config.yaml', 'r') as f:\n conf = yaml.load(f)\nBATCH_SIZE = conf[\"MODEL\"][\"BATCH_SIZE\"]\nMAX_EPOCHS = conf[\"MODEL\"][\"MAX_EPOCHS\"]\nWORD2VEC_MODEL = conf[\"EMBEDDING\"][\"WORD2VEC_MODEL\"]", "_____no_output_____" ], [ "print('load data ...')\nX_train = np.load('data/X_train.npy')\ny_train = np.load('data/y_train.npy')\nX_test = np.load('data/X_test.npy')\ny_test = np.load('data/y_test.npy')\nX_val = np.load('data/X_val.npy')\ny_val = np.load('data/y_val.npy')\n\nwith open('data/word_index.pkl', 'rb') as f:\n word_index = pickle.load(f)", "_____no_output_____" ], [ "# load embedding\nfrom gensim import models\nword2vec_model = models.KeyedVectors.load_word2vec_format(WORD2VEC_MODEL, binary=True)\nembeddings_index, embedding_dim = get_embeddings_index(word2vec_model)\nembedding_layer = get_embedding_layer(word_index, embeddings_index, embedding_dim, True)\nword2vec_model = None", "_____no_output_____" ] ], [ [ "# train model", "_____no_output_____" ] ], [ [ "lstm = get_model(2, embedding_layer)", "_____no_output_____" ], [ "callback = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=2)\nlstm.compile(\n loss='binary_crossentropy',\n optimizer=tf.keras.optimizers.Adam(0.0001),\n metrics=['accuracy'])", "_____no_output_____" ], [ "lstm.summary()", "_____no_output_____" ], [ "# train model\nhistory = lstm.fit(\n X_train, y_train,\n batch_size=BATCH_SIZE,\n epochs=MAX_EPOCHS,\n callbacks=[callback],\n validation_data=(X_test, y_test)\n)", "_____no_output_____" ] ], [ [ "# history", "_____no_output_____" ] ], [ [ "from matplotlib import pyplot as plt\n# \"Accuracy\"\nplt.plot(history.history['acc'])\nplt.plot(history.history['val_acc'])\nplt.title('model accuracy')\nplt.ylabel('accuracy')\nplt.xlabel('epoch')\nplt.legend(['train', 'validation'], loc='upper left')\nplt.show()\n# \"Loss\"\nplt.plot(history.history['loss'])\nplt.plot(history.history['val_loss'])\nplt.title('model loss')\nplt.ylabel('loss')\nplt.xlabel('epoch')\nplt.legend(['train', 'validation'], loc='upper left')\nplt.show()", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "%matplotlib inline\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport math\nimport os\nimport random\nimport pickle\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.ensemble import RandomForestRegressor\nfrom sklearn.ensemble import GradientBoostingRegressor\nfrom sklearn.ensemble import GradientBoostingClassifier\nfrom sklearn.ensemble import AdaBoostRegressor\nfrom sklearn.ensemble import AdaBoostClassifier\nfrom sklearn.tree import DecisionTreeRegressor\nfrom sklearn.metrics import mean_squared_error, r2_score\nfrom sklearn.metrics import confusion_matrix\nfrom sklearn.metrics import roc_curve, auc\nfrom sklearn import preprocessing \nfrom xgboost import XGBRegressor\nfrom datetime import datetime\nfrom bayes_opt import BayesianOptimization\nrandom.seed(1234)", "_____no_output_____" ] ], [ [ "# Define Functions", "_____no_output_____" ] ], [ [ "# Performce cross validation using xgboost\ndef xgboostcv(X, y, fold, n_estimators, lr, depth, n_jobs, gamma, min_cw, subsample, colsample):\n uid = np.unique(fold)\n model_pred = np.zeros(X.shape[0])\n model_valid_loss = np.zeros(len(uid))\n model_train_loss = np.zeros(len(uid))\n for i in uid:\n x_valid = X[fold==i]\n x_train = X[fold!=i]\n y_valid = y[fold==i]\n y_train = y[fold!=i]\n model = XGBRegressor(n_estimators=n_estimators, learning_rate=lr, \n max_depth = depth, n_jobs = n_jobs, \n gamma = gamma, min_child_weight = min_cw,\n subsample = subsample, colsample_bytree = colsample, random_state=1234)\n model.fit(x_train, y_train)\n\n pred = model.predict(x_valid)\n model_pred[fold==i] = pred\n model_valid_loss[uid==i] = mean_squared_error(y_valid, pred)\n model_train_loss[uid==i] = mean_squared_error(y_train, model.predict(x_train))\n return {'pred':model_pred, 'valid_loss':model_valid_loss, 'train_loss':model_train_loss}\n\n# Compute MSE for xgboost cross validation\ndef xgboostcv_mse(n, p, depth, g, min_cw, subsample, colsample):\n model_cv = xgboostcv(X_train, y_train, fold_train, \n int(n)*100, 10**p, int(depth), n_nodes, \n 10**g, min_cw, subsample, colsample)\n MSE = mean_squared_error(y_train, model_cv['pred'])\n return -MSE\n\n# Display model performance metrics for each cv iteration\ndef cv_performance(model, y, fold):\n uid = np.unique(fold)\n pred = np.round(model['pred'])\n y = y.reshape(-1)\n model_valid_mse = np.zeros(len(uid))\n model_valid_mae = np.zeros(len(uid))\n model_valid_r2 = np.zeros(len(uid))\n for i in uid:\n pred_i = pred[fold==i]\n y_i = y[fold==i]\n model_valid_mse[uid==i] = mean_squared_error(y_i, pred_i)\n model_valid_mae[uid==i] = np.abs(pred_i-y_i).mean()\n model_valid_r2[uid==i] = r2_score(y_i, pred_i)\n \n results = pd.DataFrame(0, index=uid, \n columns=['valid_mse', 'valid_mae', 'valid_r2', \n 'valid_loss', 'train_loss'])\n results['valid_mse'] = model_valid_mse\n results['valid_mae'] = model_valid_mae\n results['valid_r2'] = model_valid_r2\n results['valid_loss'] = model['valid_loss']\n results['train_loss'] = model['train_loss']\n print(results)\n\n# Display overall model performance metrics\ndef cv_overall_performance(y, y_pred):\n overall_MSE = mean_squared_error(y, y_pred)\n overall_MAE = (np.abs(y_pred-y)).mean()\n overall_RMSE = np.sqrt(np.square(y_pred-y).mean())\n overall_R2 = r2_score(y, y_pred)\n print(\"XGB overall MSE: %0.4f\" %overall_MSE)\n print(\"XGB overall MAE: %0.4f\" %overall_MAE)\n print(\"XGB overall RMSE: %0.4f\" %overall_RMSE)\n print(\"XGB overall R^2: %0.4f\" %overall_R2) \n\n# Plot variable importance\ndef plot_importance(model, columns):\n importances = pd.Series(model.feature_importances_, index = columns).sort_values(ascending=False)\n n = len(columns)\n plt.figure(figsize=(10,15))\n plt.barh(np.arange(n)+0.5, importances)\n plt.yticks(np.arange(0.5,n+0.5), importances.index)\n plt.tick_params(axis='both', which='major', labelsize=22)\n plt.ylim([0,n])\n plt.gca().invert_yaxis()\n plt.savefig('variable_importance.png', dpi = 150)\n\n# Save xgboost model\ndef save(obj, path):\n pkl_fl = open(path, 'wb')\n pickle.dump(obj, pkl_fl)\n pkl_fl.close()\n\n# Load xgboost model\ndef load(path):\n f = open(path, 'rb')\n obj = pickle.load(f)\n f.close()\n return(obj)", "_____no_output_____" ] ], [ [ "# Parameter Values", "_____no_output_____" ] ], [ [ "# Set a few values\nvalidation_only = False # Whether test model with test data\nn_nodes = 96 # Number of computing nodes used for hyperparamter tunning\ntrained = False # If a trained model exits\ncols_drop = ['StationId', 'Date', 'PenRate', 'NumberOfLanes', 'Dir', 'FC', 'Month'] # Columns to be dropped\nif trained:\n params = load('params.dat')\n xgb_cv = load('xgb_cv.dat')\n xgb = load('xgb.dat')", "_____no_output_____" ] ], [ [ "# Read Data", "_____no_output_____" ] ], [ [ "if validation_only:\n raw_data_train = pd.read_csv(\"final_train_data.csv\")\n data = raw_data_train.drop(cols_drop, axis=1)\n if 'Dir' in data.columns:\n data[['Dir']] = data[['Dir']].astype('category')\n one_hot = pd.get_dummies(data[['Dir']])\n data = data.drop(['Dir'], axis = 1)\n data = data.join(one_hot)\n if 'FC' in data.columns:\n data[['FC']] = data[['FC']].astype('category')\n one_hot = pd.get_dummies(data[['FC']])\n data = data.drop(['FC'], axis = 1)\n data = data.join(one_hot)\n week_dict = {\"DayOfWeek\": {'Monday': 1, 'Tuesday': 2, 'Wednesday': 3, 'Thursday': 4, \n 'Friday': 5, 'Saturday': 6, 'Sunday': 7}}\n data = data.replace(week_dict)\n\n X = data.drop(['Volume', 'fold'], axis=1)\n X_col = X.columns\n y = data[['Volume']]\n\n fold_train = data[['fold']].values.reshape(-1)\n X_train = X.values\n y_train = y.values\n \nelse:\n raw_data_train = pd.read_csv(\"final_train_data.csv\")\n raw_data_test = pd.read_csv(\"final_test_data.csv\")\n raw_data_test1 = pd.DataFrame(np.concatenate((raw_data_test.values, np.zeros(raw_data_test.shape[0]).reshape(-1, 1)), axis=1),\n columns = raw_data_test.columns.append(pd.Index(['fold'])))\n raw_data = pd.DataFrame(np.concatenate((raw_data_train.values, raw_data_test1.values), axis=0), \n columns = raw_data_train.columns)\n\n data = raw_data.drop(cols_drop, axis=1)\n if 'Dir' in data.columns:\n data[['Dir']] = data[['Dir']].astype('category')\n one_hot = pd.get_dummies(data[['Dir']])\n data = data.drop(['Dir'], axis = 1)\n data = data.join(one_hot)\n if 'FC' in data.columns:\n data[['FC']] = data[['FC']].astype('category')\n one_hot = pd.get_dummies(data[['FC']])\n data = data.drop(['FC'], axis = 1)\n data = data.join(one_hot)\n week_dict = {\"DayOfWeek\": {'Monday': 1, 'Tuesday': 2, 'Wednesday': 3, 'Thursday': 4, \n 'Friday': 5, 'Saturday': 6, 'Sunday': 7}}\n data = data.replace(week_dict)\n\n X = data.drop(['Volume'], axis=1)\n y = data[['Volume']]\n\n X_train = X.loc[X.fold!=0, :]\n fold_train = X_train[['fold']].values.reshape(-1)\n X_col = X_train.drop(['fold'], axis = 1).columns\n X_train = X_train.drop(['fold'], axis = 1).values\n y_train = y.loc[X.fold!=0, :].values\n\n X_test = X.loc[X.fold==0, :]\n X_test = X_test.drop(['fold'], axis = 1).values\n y_test = y.loc[X.fold==0, :].values", "_____no_output_____" ], [ "X_col", "_____no_output_____" ], [ "# Explain variable names\nX_name_dict = {'Temp': 'Temperature', 'WindSp': 'Wind Speed', 'Precip': 'Precipitation', 'Snow': 'Snow', \n 'Long': 'Longitude', 'Lat': 'Latitude', 'NumberOfLanes': 'Number of Lanes', 'SpeedLimit': 'Speed Limit', \n 'FRC': 'TomTom FRC', 'DayOfWeek': 'Day of Week', 'Month': 'Month', 'Hour': 'Hour', \n 'AvgSp': 'Average Speed', 'ProbeCount': 'Probe Count', 'Dir_E': 'Direction(East)', \n 'Dir_N': 'Direction(North)', 'Dir_S': 'Direction(South)', 'Dir_W': 'Direction(West)', \n 'FC_3R': 'FHWA FC(3R)', 'FC_3U': 'FHWA FC(3U)', 'FC_4R': 'FHWA FC(4R)', 'FC_4U': 'FHWA FC(4U)', \n 'FC_5R': 'FHWA FC(5R)', 'FC_5U': 'FHWA FC(5U)', 'FC_7R': 'FHWA FC(7R)', 'FC_7U': 'FHWA FC(7U)'}", "_____no_output_____" ], [ "data.head()", "_____no_output_____" ], [ "X_train.shape", "_____no_output_____" ], [ "if validation_only == False:\n print(X_test.shape)", "_____no_output_____" ] ], [ [ "# Cross Validation & Hyperparameter Optimization", "_____no_output_____" ] ], [ [ "# Set hyperparameter ranges for Bayesian optimization \nxgboostBO = BayesianOptimization(xgboostcv_mse,\n {'n': (1, 10),\n 'p': (-4, 0),\n 'depth': (2, 10),\n 'g': (-3, 0),\n 'min_cw': (1, 10), \n 'subsample': (0.5, 1), \n 'colsample': (0.5, 1)\n })", "_____no_output_____" ], [ "# Use Bayesian optimization to tune hyperparameters\nimport time\nstart_time = time.time()\nxgboostBO.maximize(init_points=10, n_iter = 50)\nprint('-'*53)\nprint('Final Results')\nprint('XGBOOST: %f' % xgboostBO.max['target'])\nprint(\"--- %s seconds ---\" % (time.time() - start_time))", "_____no_output_____" ], [ "# Save the hyperparameters the yield the highest model performance\nparams = xgboostBO.max['params']\nsave(params, 'params.dat')\nparams", "_____no_output_____" ], [ "# Perform cross validation using the optimal hyperparameters\nxgb_cv = xgboostcv(X_train, y_train, fold_train, int(params['n'])*100, \n 10**params['p'], int(params['depth']), n_nodes, \n 10**params['g'], params['min_cw'], params['subsample'], params['colsample'])", "_____no_output_____" ], [ "# Display cv results for each iteration\ncv_performance(xgb_cv, y_train, fold_train)", "_____no_output_____" ], [ "# Display overall cv results\ncv_pred = xgb_cv['pred']\ncv_pred[cv_pred<0] = 0\ncv_overall_performance(y_train.reshape(-1), cv_pred)", "_____no_output_____" ], [ "# Save the cv results\nsave(xgb_cv, 'xgb_cv.dat')", "_____no_output_____" ] ], [ [ "# Model Test", "_____no_output_____" ] ], [ [ "# Build a xgboost using all the training data with the optimal hyperparameter\nxgb = XGBRegressor(n_estimators=int(params['n'])*100, learning_rate=10**params['p'], max_depth = int(params['depth']), \n n_jobs = n_nodes, gamma = 10**params['g'], min_child_weight = params['min_cw'], \n subsample = params['subsample'], colsample_bytree = params['colsample'], random_state=1234)\nxgb.fit(X_train, y_train)", "_____no_output_____" ], [ "# Test the trained model with test data\nif validation_only == False:\n y_pred = xgb.predict(X_test)\n y_pred[y_pred<0] = 0\n cv_overall_performance(y_test.reshape(-1), y_pred)", "_____no_output_____" ], [ "# Plot variable importance\ncol_names = [X_name_dict[i] for i in X_col]\nplot_importance(xgb, col_names)", "_____no_output_____" ], [ "# Save the trained xgboost model\nsave(xgb, 'xgb.dat')", "_____no_output_____" ], [ "# Produce cross validation estimates or estimates for test data\ntrain_data_pred = pd.DataFrame(np.concatenate((raw_data_train.values, cv_pred.reshape(-1, 1)), axis=1),\n columns = raw_data_train.columns.append(pd.Index(['PredVolume'])))\ntrain_data_pred.to_csv('train_data_pred.csv', index = False)\n\nif validation_only == False:\n test_data_pred = pd.DataFrame(np.concatenate((raw_data_test.values, y_pred.reshape(-1, 1)), axis=1),\n columns = raw_data_test.columns.append(pd.Index(['PredVolume'])))\n test_data_pred.to_csv('test_data_pred.csv', index = False)\n", "_____no_output_____" ] ], [ [ "# Plot Estimations vs. Observations", "_____no_output_____" ] ], [ [ "# Prepare data to plot estimated and observed values\nif validation_only:\n if trained:\n plot_df = pd.read_csv(\"train_data_pred.csv\")\n else:\n plot_df = train_data_pred\nelse:\n if trained:\n plot_df = pd.read_csv(\"test_data_pred.csv\")\n else:\n plot_df = test_data_pred \nplot_df = plot_df.sort_values(by=['StationId', 'Date', 'Dir', 'Hour'])\nplot_df = plot_df.set_index(pd.Index(range(plot_df.shape[0])))", "_____no_output_____" ], [ "# Define a function to plot estimated and observed values for a day\ndef plot_daily_estimate(frc):\n indices = plot_df.index[(plot_df.FRC == frc) & (plot_df.Hour == 0)].tolist()\n from_index = np.random.choice(indices, 1)[0]\n to_index = from_index + 23\n plot_df_sub = plot_df.loc[from_index:to_index, :]\n time = pd.date_range(plot_df_sub.Date.iloc[0] + ' 00:00:00', periods=24, freq='H')\n plt.figure(figsize=(20,10))\n plt.plot(time, plot_df_sub.PredVolume, 'b-', label='XGBoost', lw=2)\n plt.plot(time, plot_df_sub.Volume, 'r--', label='Observed', lw=3)\n plt.tick_params(axis='both', which='major', labelsize=24)\n plt.ylabel('Volume (vehs/hr)', fontsize=24)\n plt.xlabel(\"Time\", fontsize=24)\n plt.legend(loc='upper left', shadow=True, fontsize=24)\n plt.title('Station ID: {0}, MAE={1}, FRC = {2}'.format(\n plot_df_sub.StationId.iloc[0],\n round(np.abs(plot_df_sub.PredVolume-plot_df_sub.Volume).mean()),\n plot_df_sub.FRC.iloc[0]), fontsize=40)\n plt.savefig('frc_{0}.png'.format(frc), dpi = 150)\n return(plot_df_sub)", "_____no_output_____" ], [ "# Define a function to plot estimated and observed values for a week\ndef plot_weekly_estimate(frc):\n indices = plot_df.index[(plot_df.FRC == frc) & (plot_df.Hour == 0) & (plot_df.DayOfWeek == 'Monday')].tolist()\n from_index = np.random.choice(indices, 1)[0]\n to_index = from_index + 24*7-1\n plot_df_sub = plot_df.loc[from_index:to_index, :]\n time = pd.date_range(plot_df_sub.Date.iloc[0] + ' 00:00:00', periods=24*7, freq='H')\n plt.figure(figsize=(20,10))\n plt.plot(time, plot_df_sub.PredVolume, 'b-', label='XGBoost', lw=2)\n plt.plot(time, plot_df_sub.Volume, 'r--', label='Observed', lw=3)\n plt.tick_params(axis='both', which='major', labelsize=24)\n plt.ylabel('Volume (vehs/hr)', fontsize=24)\n plt.xlabel(\"Time\", fontsize=24)\n plt.legend(loc='upper left', shadow=True, fontsize=24)\n plt.title('Station ID: {0}, MAE={1}, FRC = {2}'.format(\n plot_df_sub.StationId.iloc[0],\n round(np.abs(plot_df_sub.PredVolume-plot_df_sub.Volume).mean()),\n plot_df_sub.FRC.iloc[0]), fontsize=40)\n plt.savefig('frc_{0}.png'.format(frc), dpi = 150)\n return(plot_df_sub)", "_____no_output_____" ], [ "# Plot estimated and observed values for a day\nfrc2_daily_plot = plot_daily_estimate(2)\nsave(frc2_daily_plot, 'frc2_daily_plot.dat')", "_____no_output_____" ], [ "# Plot estimated and observed values for a week\nfrc3_weekly_plot = plot_weekly_estimate(3)\nsave(frc3_weekly_plot, 'frc3_weekly_plot.dat')", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "First import the \"datavis\" module", "_____no_output_____" ] ], [ [ "import sys\nsys.path.append('..')\nimport numpy as np\nimport datavis \nimport vectorized_datavis", "_____no_output_____" ], [ "def test_se_to_sd():\n \"\"\"\n Test that the value returned is a float value\n \"\"\"\n sdev = datavis.se_to_sd(0.5, 1000)\n assert isinstance(sdev, float),\\\n \"Returned data type is not a float number\"", "_____no_output_____" ], [ "test_se_to_sd() ", "_____no_output_____" ], [ "def test_ci_to_sd():\n \"\"\"\n Test that the value returned is a float value\n \"\"\"\n sdev = datavis.ci_to_sd(0.2, 0.4) \n assert isinstance(sdev, float),\\\n \"Returned data type is not a float number\"", "_____no_output_____" ], [ "test_ci_to_sd() ", "_____no_output_____" ], [ "def test_datagen():\n \"\"\"\n Test that the data returned is a numpy.ndarray\n \"\"\"\n randdata = datavis.datagen(25, 0.2, 0.4)\n assert isinstance(randdata, np.ndarray),\\\n \"Returned data type is not a numpy.ndarray\"", "_____no_output_____" ], [ "test_datagen() ", "_____no_output_____" ], [ "def test_correctdatatype():\n \"\"\"\n Test that the statistical parameters returned are float numbers\n \"\"\"\n fmean, fsdev, fserror, fuci, flci = datavis.correctdatatype(0.2, 0.4)\n assert isinstance(fmean, float),\\\n \"Returned data type is not a float number\"", "_____no_output_____" ], [ "test_correctdatatype() ", "_____no_output_____" ], [ "def test_compounddata():\n \"\"\"\n Test that the data returned are numpy.ndarrays\n \"\"\"\n datagenerated1, datagenerated2, datagenerated3 = \\\n datavis.compounddata\\\n (mean1=24.12,sdev1=3.87,mean2=24.43,sdev2=3.94,mean3=24.82,sdev3=3.95,size=1000) \n assert isinstance(datagenerated1, np.ndarray),\\\n \"Returned data are not numpy.ndarrays\"", "_____no_output_____" ], [ "test_compounddata() ", "_____no_output_____" ], [ "def test_databinning():\n \"\"\"\n Test that the data returned are numpy.ndarrays\n \"\"\"\n datagenerated1, datagenerated2, datagenerated3 = \\\n datavis.compounddata\\\n (mean1=24.12,sdev1=3.87,mean2=24.43,sdev2=3.94,mean3=24.82,sdev3=3.95,size=1000)\n bins = np.linspace(10,40,num=30)\n yhist1, yhist2, yhist3 = \\\n datavis.databinning\\\n (datagenerated1, datagenerated2, datagenerated3,bins_list=bins) \n assert isinstance(yhist1, np.ndarray),\\\n \"Returned data are not numpy.ndarrays\"", "_____no_output_____" ], [ "test_databinning() ", "_____no_output_____" ], [ "def test_pdfgen():\n \"\"\"\n Test that the data returned are numpy.ndarrays\n \"\"\"\n bins = np.linspace(10,40,num=30)\n mean1 = 24.12\n sdev1 = 3.87\n mean2 = 24.43\n sdev2 = 3.94\n mean3 = 24.82\n sdev3 = 3.95 \n pdf1, pdf2, pdf3 = datavis.pdfgen\\\n (mean1, sdev1, mean2, sdev2, mean3, sdev3, bins_list=bins)\n assert isinstance(pdf1, np.ndarray),\\\n \"Returned data are not numpy.ndarrays\"", "_____no_output_____" ], [ "test_pdfgen() ", "_____no_output_____" ], [ "def test_percent_overlap():\n \"\"\"\n Test that the data returned is a tuple\n \"\"\"\n mean1 = 24.12\n sdev1 = 3.87\n mean2 = 24.43\n sdev2 = 3.94\n mean3 = 24.82\n sdev3 = 3.95\n overlap_11_perc, overlap_12_perc, overlap_13_perc = \\\n datavis.percent_overlap\\\n (mean1, sdev1, mean2, sdev2, mean3, sdev3)\n assert isinstance\\\n (datavis.percent_overlap\\\n (mean1, sdev1, mean2, sdev2, mean3, sdev3), tuple),\\\n \"Returned data is not numpy.float64 type\"", "_____no_output_____" ], [ "test_percent_overlap() ", "_____no_output_____" ], [ "mean = np.array([24.12, 24.43, 24.82])\nsdev = np.array([3.87, 3.94, 3.95]) ", "_____no_output_____" ], [ "vectorized_datavis.compounddata(mean, sdev) ", "_____no_output_____" ], [ "def test_compounddata():\n \"\"\"\n Test that the data returned are numpy.ndarrays\n \"\"\"\n mean = np.array([24.12, 24.43, 24.82])\n sdev = np.array([3.87, 3.94, 3.95]) \n datagenerated = vectorized_datavis.compounddata(mean, sdev) \n assert isinstance(datagenerated, np.ndarray),\\\n \"Returned data are not numpy.ndarrays\"", "_____no_output_____" ], [ "test_compounddata() ", "_____no_output_____" ], [ "def test_databinning():\n \"\"\"\n Test that the data returned are numpy.ndarrays\n \"\"\"\n mean = np.array([24.12, 24.43, 24.82])\n sdev = np.array([3.87, 3.94, 3.95]) \n datagenerated = vectorized_datavis.compounddata(mean, sdev)\n bins = np.linspace(10, 40, num=30)\n yhist = vectorized_datavis.databinning(datagenerated, bins_list=bins) \n assert isinstance(yhist, np.ndarray),\\\n \"Returned data are not numpy.ndarrays\" ", "_____no_output_____" ], [ "test_databinning()", "_____no_output_____" ], [ "def test_pdfgen():\n \"\"\"\n Test that the data returned are numpy.ndarrays\n \"\"\"\n bins = np.linspace(10,40,num=30)\n mean = np.array([24.12, 24.43, 24.82])\n sdev = np.array([3.87, 3.94, 3.95]) \n pdf = vectorized_datavis.pdfgen(mean, sdev, bins_list=bins)\n assert isinstance(pdf, np.ndarray),\\\n \"Returned data are not numpy.ndarrays\"", "_____no_output_____" ], [ "test_pdfgen() ", "_____no_output_____" ], [ "def test_percent_overlap():\n \"\"\"\n Test that the data returned is a numpy.ndarray\n \"\"\"\n mean = np.array([24.12, 24.43, 24.82])\n sdev = np.array([3.87, 3.94, 3.95]) \n overlap_perc_1w = vectorized_datavis.percent_overlap\\\n (mean, sdev)\n assert isinstance(overlap_perc_1w, np.ndarray),\\\n \"Returned data is not a numpy.ndarray\"", "_____no_output_____" ], [ "test_percent_overlap() ", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# IDS Instruction: Regression\n(Lisa Mannel)", "_____no_output_____" ], [ "## Simple linear regression", "_____no_output_____" ], [ "First we import the packages necessary fo this instruction:", "_____no_output_____" ] ], [ [ "import numpy as np\nimport matplotlib.pyplot as plt\nimport pandas as pd\nfrom sklearn.linear_model import LinearRegression\nfrom sklearn.metrics import mean_squared_error, mean_absolute_error", "_____no_output_____" ] ], [ [ "Consider the data set \"df\" with feature variables \"x\" and \"y\" given below.", "_____no_output_____" ] ], [ [ "df1 = pd.DataFrame({'x': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], 'y': [1, 3, 2, 5, 7, 8, 8, 9, 10, 12]})\nprint(df1)", " x y\n0 0 1\n1 1 3\n2 2 2\n3 3 5\n4 4 7\n5 5 8\n6 6 8\n7 7 9\n8 8 10\n9 9 12\n" ] ], [ [ "To get a first impression of the given data, let's have a look at its scatter plot:", "_____no_output_____" ] ], [ [ "plt.scatter(df1.x, df1.y, color = \"y\", marker = \"o\", s = 40)\nplt.xlabel('x') \nplt.ylabel('y')\nplt.title('first overview of the data')\nplt.show()", "_____no_output_____" ] ], [ [ "We can already see a linear correlation between x and y. Assume the feature x to be descriptive, while y is our target feature. We want a linear function, y=ax+b, that predicts y as accurately as possible based on x. To achieve this goal we use linear regression from the sklearn package.", "_____no_output_____" ] ], [ [ "#define the set of descriptive features (in this case only 'x' is in that set) and the target feature (in this case 'y')\ndescriptiveFeatures1=df1[['x']]\ntargetFeature1=df1['y']\n\n#define the classifier\nclassifier = LinearRegression()\n#train the classifier\nmodel1 = classifier.fit(descriptiveFeatures1, targetFeature1)", "_____no_output_____" ] ], [ [ "Now we can use the classifier to predict y. We print the predictions as well as the coefficient and bias (*intercept*) of the linear function.", "_____no_output_____" ] ], [ [ "#use the classifier to make prediction\ntargetFeature1_predict = classifier.predict(descriptiveFeatures1)\nprint(targetFeature1_predict)\n#print coefficient and intercept\nprint('Coefficients: \\n', classifier.coef_)\nprint('Intercept: \\n', classifier.intercept_)", "[ 1.23636364 2.40606061 3.57575758 4.74545455 5.91515152 7.08484848\n 8.25454545 9.42424242 10.59393939 11.76363636]\nCoefficients: \n [1.16969697]\nIntercept: \n 1.2363636363636399\n" ] ], [ [ "Let's visualize our regression function with the scatterplot showing the original data set. Herefore, we use the predicted values.", "_____no_output_____" ] ], [ [ "#visualize data points\nplt.scatter(df1.x, df1.y, color = \"y\", marker = \"o\", s = 40) \n#visualize regression function\nplt.plot(descriptiveFeatures1, targetFeature1_predict, color = \"g\") \nplt.xlabel('x') \nplt.ylabel('y') \nplt.title('the data and the regression function')\nplt.show()", "_____no_output_____" ] ], [ [ "### <span style=\"color:green\"> Now it is your turn. </span> Build a simple linear regression for the data below. Use col1 as descriptive feature and col2 as target feature. Also plot your results.", "_____no_output_____" ] ], [ [ "df2 = pd.DataFrame({'col1': [770, 677, 428, 410, 371, 504, 1136, 695, 551, 550], 'col2': [54, 47, 28, 38, 29, 38, 80, 52, 45, 40]})\n#Your turn\n", "_____no_output_____" ] ], [ [ "### Evaluation\n\nUsually, the model and its predictions is not sufficient. In the following we want to evaluate our classifiers. \n\nLet's start by computing their error. The sklearn.metrics package contains several errors such as\n\n* Mean squared error\n* Mean absolute error\n* Mean squared log error\n* Median absolute error\n", "_____no_output_____" ] ], [ [ "#computing the squared error of the first model\nprint(\"Mean squared error model 1: %.2f\" % mean_squared_error(targetFeature1, targetFeature1_predict))", "Mean squared error model 1: 0.56\n" ] ], [ [ "We can also visualize the errors:", "_____no_output_____" ] ], [ [ "plt.scatter(targetFeature1_predict, (targetFeature1 - targetFeature1_predict) ** 2, color = \"blue\", s = 10,) \n \n## plotting line to visualize zero error \nplt.hlines(y = 0, xmin = 0, xmax = 15, linewidth = 2) \n \n## plot title \nplt.title(\"Squared errors Model 1\") \n \n## function to show plot \nplt.show() ", "_____no_output_____" ] ], [ [ "### <span style=\"color:green\"> Now it is your turn. </span> Compute the mean squared error and visualize the squared errors. Play around using different error metrics.", "_____no_output_____" ] ], [ [ "#Your turn\n", "_____no_output_____" ] ], [ [ "## Handling multiple descriptive features at once - Multiple linear regression\nIn most cases, we will have more than one descriptive feature . As an example we use an example data set of the scikit package. The dataset describes housing prices in Boston based on several attributes. Note, in this format the data is already split into descriptive features and a target feature.", "_____no_output_____" ] ], [ [ "from sklearn import datasets ## imports datasets from scikit-learn\ndf3 = datasets.load_boston()\n\n#The sklearn package provides the data splitted into a set of descriptive features and a target feature.\n#We can easily transform this format into the pandas data frame as used above.\ndescriptiveFeatures3 = pd.DataFrame(df3.data, columns=df3.feature_names)\ntargetFeature3 = pd.DataFrame(df3.target, columns=['target'])\nprint('Descriptive features:')\nprint(descriptiveFeatures3.head())\nprint('Target feature:')\nprint(targetFeature3.head())", "Descriptive features:\n CRIM ZN INDUS CHAS NOX RM AGE DIS RAD TAX \\\n0 0.00632 18.0 2.31 0.0 0.538 6.575 65.2 4.0900 1.0 296.0 \n1 0.02731 0.0 7.07 0.0 0.469 6.421 78.9 4.9671 2.0 242.0 \n2 0.02729 0.0 7.07 0.0 0.469 7.185 61.1 4.9671 2.0 242.0 \n3 0.03237 0.0 2.18 0.0 0.458 6.998 45.8 6.0622 3.0 222.0 \n4 0.06905 0.0 2.18 0.0 0.458 7.147 54.2 6.0622 3.0 222.0 \n\n PTRATIO B LSTAT \n0 15.3 396.90 4.98 \n1 17.8 396.90 9.14 \n2 17.8 392.83 4.03 \n3 18.7 394.63 2.94 \n4 18.7 396.90 5.33 \nTarget feature:\n target\n0 24.0\n1 21.6\n2 34.7\n3 33.4\n4 36.2\n" ] ], [ [ "To predict the housing price we will use a Multiple Linear Regression model. In Python this is very straightforward: we use the same function as for simple linear regression, but our set of descriptive features now contains more than one element (see above).", "_____no_output_____" ] ], [ [ "classifier = LinearRegression()\nmodel3 = classifier.fit(descriptiveFeatures3,targetFeature3)\n\ntargetFeature3_predict = classifier.predict(descriptiveFeatures3)\nprint('Coefficients: \\n', classifier.coef_)\nprint('Intercept: \\n', classifier.intercept_)\nprint(\"Mean squared error: %.2f\" % mean_squared_error(targetFeature3, targetFeature3_predict))", "Coefficients: \n [[-1.08011358e-01 4.64204584e-02 2.05586264e-02 2.68673382e+00\n -1.77666112e+01 3.80986521e+00 6.92224640e-04 -1.47556685e+00\n 3.06049479e-01 -1.23345939e-02 -9.52747232e-01 9.31168327e-03\n -5.24758378e-01]]\nIntercept: \n [36.45948839]\nMean squared error: 21.89\n" ] ], [ [ "As you can see above, we have a coefficient for each descriptive feature.", "_____no_output_____" ], [ "## Handling categorical descriptive features\nSo far we always encountered numerical dscriptive features, but data sets can also contain categorical attributes. The regression function can only handle numerical input. There are several ways to tranform our categorical data to numerical data (for example using one-hot encoding as explained in the lecture: we introduce a 0/1 feature for every possible value of our categorical attribute). For adequate data, another possibility is to replace each categorical value by a numerical value and adding an ordering with it. \n\nPopular possibilities to achieve this transformation are\n\n* the get_dummies function of pandas\n* the OneHotEncoder of scikit\n* the LabelEncoder of scikit\n\nAfter encoding the attributes we can apply our regular regression function.", "_____no_output_____" ] ], [ [ "#example using pandas\ndf4 = pd.DataFrame({'A':['a','b','c'],'B':['c','b','a'] })\none_hot_pd = pd.get_dummies(df4)\none_hot_pd", "_____no_output_____" ], [ "#example using scikit\nfrom sklearn.preprocessing import LabelEncoder, OneHotEncoder\n\n#apply the one hot encoder\nencoder = OneHotEncoder(categories='auto')\nencoder.fit(df4)\ndf4_OneHot = encoder.transform(df4).toarray()\nprint('Transformed by One-hot Encoding: ')\nprint(df4_OneHot)\n\n# encode labels with value between 0 and n_classes-1\nencoder = LabelEncoder()\ndf4_LE = df4.apply(encoder.fit_transform)\nprint('Replacing categories by numerical labels: ')\nprint(df4_LE.head())", "Transformed by One-hot Encoding: \n[[1. 0. 0. 0. 0. 1.]\n [0. 1. 0. 0. 1. 0.]\n [0. 0. 1. 1. 0. 0.]]\nReplacing categories by numerical labels: \n A B\n0 0 2\n1 1 1\n2 2 0\n" ] ], [ [ "### <span style=\"color:green\"> Now it is your turn. </span> Perform linear regression using the data set given below. Don't forget to transform your categorical descriptive features. The rental price attribute represents the target variable. ", "_____no_output_____" ] ], [ [ "df5 = pd.DataFrame({'Size':[500,550,620,630,665],'Floor':[4,7,9,5,8], 'Energy rating':['C', 'A', 'A', 'B', 'C'], 'Rental price': [320,380,400,390,385] })\n#Your turn\n", "_____no_output_____" ] ], [ [ "## Predicting a categorical target value - Logistic regression", "_____no_output_____" ], [ "We might also encounter data sets where our target feature is categorical. Here we don't transform them into numerical values, but insetad we use a logistic regression function. Luckily, sklearn provides us with a suitable function that is similar to the linear equivalent. Similar to linear regression, we can compute logistic regression on a single descriptive variable as well as on multiple variables.", "_____no_output_____" ] ], [ [ "# Importing the dataset\niris = pd.read_csv('iris.csv')\n\nprint('First look at the data set: ')\nprint(iris.head())\n\n#defining the descriptive and target features\ndescriptiveFeatures_iris = iris[['sepal_length']] #we only use the attribute 'sepal_length' in this example\ntargetFeature_iris = iris['species'] #we want to predict the 'species' of iris\n\nfrom sklearn.linear_model import LogisticRegression\nclassifier = LogisticRegression(solver = 'liblinear', multi_class = 'ovr')\nclassifier.fit(descriptiveFeatures_iris, targetFeature_iris)\n\ntargetFeature_iris_pred = classifier.predict(descriptiveFeatures_iris)\n\nprint('Coefficients: \\n', classifier.coef_)\nprint('Intercept: \\n', classifier.intercept_)", "First look at the data set: \n sepal_length sepal_width petal_length petal_width species\n0 5.1 3.5 1.4 0.2 setosa\n1 4.9 3.0 1.4 0.2 setosa\n2 4.7 3.2 1.3 0.2 setosa\n3 4.6 3.1 1.5 0.2 setosa\n4 5.0 3.6 1.4 0.2 setosa\nCoefficients: \n [[-0.86959145]\n [ 0.01223362]\n [ 0.57972675]]\nIntercept: \n [ 4.16186636 -0.74244291 -3.9921824 ]\n" ] ], [ [ "### <span style=\"color:green\"> Now it is your turn. </span> In the example above we only used the first attribute as descriptive variable. Change the example such that all available attributes are used.", "_____no_output_____" ] ], [ [ "#Your turn", "_____no_output_____" ] ], [ [ "Note, that the regression classifier (both logistic and non-logistic) can be tweaked using several parameters. This includes, but is not limited to, non-linear regression. Check out the documentation for details and feel free to play around!", "_____no_output_____" ], [ "# Support Vector Machines", "_____no_output_____" ], [ "Aside from regression models, the sklearn package also provides us with a function for training support vector machines. Looking at the example below we see that they can be trained in similar ways. We still use the iris data set for illustration.", "_____no_output_____" ] ], [ [ "from sklearn.svm import SVC\n\n#define descriptive and target features as before\ndescriptiveFeatures_iris = iris[['sepal_length', 'sepal_width', 'petal_length', 'petal_width']]\ntargetFeature_iris = iris['species']\n\n#this time, we train an SVM classifier\nclassifier = SVC(C=1, kernel='linear', gamma = 'auto')\nclassifier.fit(descriptiveFeatures_iris, targetFeature_iris)\n\ntargetFeature_iris_predict = classifier.predict(descriptiveFeatures_iris)\ntargetFeature_iris_predict[0:5] #show the first 5 predicted values", "_____no_output_____" ] ], [ [ "As explained in the lecture, a support vector machine is defined by its support vectors. In the sklearn package we can access them and their properties very easily:\n\n* support_: indicies of support vectors\n* support_vectors_: the support vectors\n* n_support_: the number of support vectors for each class", "_____no_output_____" ] ], [ [ "print('Indicies of support vectors:')\nprint(classifier.support_)\n\nprint('The support vectors:')\nprint(classifier.support_vectors_)\n\nprint('The number of support vectors for each class:')\nprint(classifier.n_support_)", "Indicies of support vectors:\n[ 23 24 41 52 56 63 66 68 70 72 76 77 83 84 98 106 110 119\n 123 126 127 129 133 138 146 147 149]\nThe support vectors:\n[[5.1 3.3 1.7 0.5]\n [4.8 3.4 1.9 0.2]\n [4.5 2.3 1.3 0.3]\n [6.9 3.1 4.9 1.5]\n [6.3 3.3 4.7 1.6]\n [6.1 2.9 4.7 1.4]\n [5.6 3. 4.5 1.5]\n [6.2 2.2 4.5 1.5]\n [5.9 3.2 4.8 1.8]\n [6.3 2.5 4.9 1.5]\n [6.8 2.8 4.8 1.4]\n [6.7 3. 5. 1.7]\n [6. 2.7 5.1 1.6]\n [5.4 3. 4.5 1.5]\n [5.1 2.5 3. 1.1]\n [4.9 2.5 4.5 1.7]\n [6.5 3.2 5.1 2. ]\n [6. 2.2 5. 1.5]\n [6.3 2.7 4.9 1.8]\n [6.2 2.8 4.8 1.8]\n [6.1 3. 4.9 1.8]\n [7.2 3. 5.8 1.6]\n [6.3 2.8 5.1 1.5]\n [6. 3. 4.8 1.8]\n [6.3 2.5 5. 1.9]\n [6.5 3. 5.2 2. ]\n [5.9 3. 5.1 1.8]]\nThe number of support vectors for each class:\n[ 3 12 12]\n" ] ], [ [ "We can also calculate the distance of the data points to the separating hyperplane by using the decision_function(X) method. Score(X,y) calculates the mean accuracy of the classification. The classification report shows metrics such as precision, recall, f1-score and support. You will learn more about these quality metrics in a few lectures.", "_____no_output_____" ] ], [ [ "from sklearn.metrics import classification_report\nclassifier.decision_function(descriptiveFeatures_iris)\nprint('Accuracy: \\n', classifier.score(descriptiveFeatures_iris,targetFeature_iris))\nprint('Classification report: \\n')\nprint(classification_report(targetFeature_iris, targetFeature_iris_predict)) ", "Accuracy: \n 0.9933333333333333\nClassification report: \n\n precision recall f1-score support\n\n setosa 1.00 1.00 1.00 50\n versicolor 1.00 0.98 0.99 50\n virginica 0.98 1.00 0.99 50\n\n accuracy 0.99 150\n macro avg 0.99 0.99 0.99 150\nweighted avg 0.99 0.99 0.99 150\n\n" ] ], [ [ "The SVC has many parameters. In the lecture you learned about the concept of kernels. Scikit gives you the opportunity to try different kernel functions.\nFurthermore, the parameter C tells the SVM optimization problem how much you want to avoid misclassifying each training example. ", "_____no_output_____" ], [ "On the scikit website you can find more information about the available kernels etc. http://scikit-learn.org/stable/modules/generated/sklearn.svm.SVC.html", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "import pandas as pd", "_____no_output_____" ], [ "products = pd.read_csv('olist_products_dataset.csv')\nproducts.head()", "_____no_output_____" ], [ "# find the most common product color\nproducts.product_category_name.value_counts().idxmax()", "_____no_output_____" ], [ "orders = pd.read_csv('olist_orders_dataset.csv')\norders.head()", "_____no_output_____" ], [ "total = pd.merge(transacitions, product, on='product_id')", "_____no_output_____" ], [ "orders['date'] = pd.to_datetime(orders.order_purchase_timestamp)\n\n# red or not\ndef status(x):\n if x == 'delivered':\n return 1\n else:\n return 0\n\norders['status'] = orders.order_status.apply(status)\n\norders.head()", "_____no_output_____" ], [ "print(orders.date.max())\nprint(orders.date.min())\nprint(orders.date.apply(['max','min']))", "2018-10-17 17:30:18\n2016-09-04 21:15:19\nmax 2018-10-17 17:30:18\nmin 2016-09-04 21:15:19\nName: date, dtype: datetime64[ns]\n" ], [ "#aggregate the data by customer and filter it to the pre-Oct-2012 time period\nagg = orders[orders['date'] < '2018-01-01 00:00:00'].groupby('customer_id')", "_____no_output_____" ], [ "# use this data to create as many predictor features\n\n# number of transcactins\nnumber_of_transactions = agg.count()['order_id']\n\n# total money spent\n## aggregate on the original transactions_data\n## transactions['total_money'] = transactions['purchased_quantity'] * transactions['unit_product_price'] \n## agg.total_money.sum()\n\n# red product\nred_or_not = agg.sum()\n\n# total purchased quantity\n\n# average price for product\n## total money / total quantity\n\n# transaction frequency", "_____no_output_____" ], [ "feature = pd.concat([number_of_transactions, red_or_not],axis = 1)\nfeature.head()", "_____no_output_____" ], [ "# create response feature\nimport numpy as np\norder2018 = orders[orders['date'] >= '2018-01-01 00:00:00']\norder2018['target'] = np.ones(order2018.shape[0],dtype=int)\ntarget = order2018.groupby('customer_id').sum().drop('status',axis = 1)\ntarget.head()", "//anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\n\nSee the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n after removing the cwd from sys.path.\n" ], [ "features = pd.merge(feature, target, how = 'left',on = 'customer_id')\nfeatures.head()", "_____no_output_____" ], [ "# change NaN to 0\nfeatures = features.fillna(0)\nfeatures.index.is_unique\nfeatures.head()", "_____no_output_____" ], [ "#save output data as a csv file\n\nfeatures.to_csv('features_data.csv')", "_____no_output_____" ], [ "data = pd.read_csv('features_data.csv',index_col = 'customer_id')\ndata.head()", "_____no_output_____" ], [ "X = data.drop('target', axis = 1)\ny = data['target']\n\n# train_test_split\nfrom sklearn.model_selection import train_test_split\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)\n\nfrom sklearn.linear_model import LogisticRegression\nlogmodel = LogisticRegression()\nlogmodel.fit(X_train,y_train)\npredictions = logmodel.predict(X_test)", "//anaconda3/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n FutureWarning)\n" ], [ "#evaluation\nfrom sklearn.metrics import classification_report\nprint(classification_report(y_test,predictions))", " precision recall f1-score support\n\n 0.0 1.00 1.00 1.00 15042\n 1.0 1.00 1.00 1.00 17774\n\n accuracy 1.00 32816\n macro avg 1.00 1.00 1.00 32816\nweighted avg 1.00 1.00 1.00 32816\n\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import scraping_class\nlogfile = 'log.txt'## name your log file.\nconnector = scraping_class.Connector(logfile)\nimport requests\nfrom bs4 import BeautifulSoup\nfrom tqdm import tqdm_notebook", "_____no_output_____" ], [ "import pandas as pd\nimport numpy as np\nimport html5lib\nimport sys\nimport pickle\nfrom tqdm import tqdm_notebook\nimport seaborn as sns", "_____no_output_____" ], [ "import matplotlib.pyplot as plt\nfrom matplotlib import pyplot", "_____no_output_____" ], [ "with open('df_final.pkl', 'rb') as f:\n df = pickle.load(f)", "_____no_output_____" ] ], [ [ "### Grade distrubtion by Faculties (used in the appendix)", "_____no_output_____" ] ], [ [ "# note: no NaN's\nprint(df['Fakultet'].unique())", "_____no_output_____" ], [ "# Number of courses by faculty\nprint(sum(df['Fakultet'] == 'Det Natur- og Biovidenskabelige Fakultet'))\nprint(sum(df['Fakultet'] == 'Det Samfundsvidenskabelige Fakultet'))\nprint(sum(df['Fakultet'] == 'Det Humanistiske Fakultet'))\nprint(sum(df['Fakultet'] == 'Det Sundhedsvidenskabelige Fakultet'))\nprint(sum(df['Fakultet'] == 'Det Juridiske Fakultet'))\nprint(sum(df['Fakultet'] == 'Det Teologiske Fakultet'))", "_____no_output_____" ], [ "#Number of grades given for each faculty. \nlist_number_of_grades_faculties = []\nfor i in tqdm_notebook(df['Fakultet'].unique()):\n df_number_grades = df[df['Fakultet'] == i]\n list_number_of_grades_faculties.append(int(sum(df_number_grades[[12, 10, 7, 4, 2, 0, -3]].sum(skipna = True))))\n\nlist_number_of_grades_faculties;", "_____no_output_____" ], [ "#Number of passing grades given for each faculty. \nlist_no_fail_number_of_grades_faculties = []\nfor i in tqdm_notebook(df['Fakultet'].unique()):\n df_number_grades = df[df['Fakultet'] == i]\n list_no_fail_number_of_grades_faculties.append(int(sum(df_number_grades[[12, 10, 7, 4, 2]].sum(skipna = True))))\n\nlist_no_fail_number_of_grades_faculties", "_____no_output_____" ] ], [ [ "### Grade distrubtion by Faculties, weighted against ECTS", "_____no_output_____" ] ], [ [ "# We need to weight the grades according to ECTS points. If we do not small courses will have the same weight as\n# bigger courses. \ndf['Weigthed_m3'] = df['Credit_edit'] * df[-3]\ndf['Weigthed_00'] = df['Credit_edit'] * df[0]\ndf['Weigthed_02'] = df['Credit_edit'] * df[2]\ndf['Weigthed_4'] = df['Credit_edit'] * df[4]\ndf['Weigthed_7'] = df['Credit_edit'] * df[7]\ndf['Weigthed_10'] = df['Credit_edit'] * df[10]\ndf['Weigthed_12'] = df['Credit_edit'] * df[12]\ndf[['Credit_edit',-3,'Weigthed_m3',0,'Weigthed_00',2,'Weigthed_02',4,'Weigthed_4',7,'Weigthed_7',10,'Weigthed_10',12,'Weigthed_12']];", "_____no_output_____" ], [ "y_ects_inner = []\ny_ects = []\nx = ['-3','00','02','4','7','10','12']\n\n# Looking at each faculty \nfor i in tqdm_notebook(df['Fakultet'].unique()):\n df_faculty = df[df['Fakultet']==i]\n \n # Using the weighted grades this time.\n for k in ['Weigthed_m3','Weigthed_00','Weigthed_02','Weigthed_4','Weigthed_7','Weigthed_10','Weigthed_12']:\n y_ects_inner.append(df_faculty[k].sum(skipna = True)) \n \n y_ects.append(y_ects_inner)\n y_ects_inner=[]", "_____no_output_____" ], [ "# calc frequencies \ny_ects_freq_inner = []\ny_ects_freq = []\n\n# running through each faculty\nfor i in range(len(y_ects)):\n \n # calc frequencies \n for q in range(len(y_ects[i])):\n y_ects_freq_inner.append(y_ects[i][q]/sum(y_ects[i]))\n\n y_ects_freq.append(y_ects_freq_inner)\n y_ects_freq_inner = []", "_____no_output_____" ], [ "# This figure is used in the analysis. It uses WEIGHTED grades\nf, ax = plt.subplots(figsize=(15,10))\n\nplt.subplot(2, 3, 1)\nplt.title(Faculty_names[0], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.30])\nplt.grid(axis ='y',zorder=0)\nplt.ylabel('Frequency',fontsize=14)\nplt.annotate('Number of grades th=0.93, edgecolor='black',zorder=3)\ngiven: '+ str(list_number_of_grades_faculties[0]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x, y_ects_freq[0], wid\nplt.subplot(2, 3, 2)\nplt.title(Faculty_names[1], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.30])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of grades given: '+ str(list_number_of_grades_faculties[1]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x, y_ects_freq[1], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 3)\nplt.title(Faculty_names[2], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.30])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of grades given: '+ str(list_number_of_grades_faculties[2]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x, y_ects_freq[2], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 4)\nplt.title(Faculty_names[3], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.30])\nplt.grid(axis ='y',zorder=0)\nplt.ylabel('Frequency',fontsize=14)\nplt.annotate('Number of grades given: '+ str(list_number_of_grades_faculties[3]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x, y_ects_freq[3], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 5)\nplt.title(Faculty_names[4], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.30])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of grades given: '+ str(list_number_of_grades_faculties[4]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x, y_ects_freq[4], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 6)\nplt.title(Faculty_names[5], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.30])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of grades given: '+ str(list_number_of_grades_faculties[5]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x, y_ects_freq[5], width=0.93, edgecolor='black',zorder=3)\n\nf.savefig('histogram_gades_split_faculty_ECTS_weight.png')", "_____no_output_____" ] ], [ [ "### Faculties, weighted against ECTS - droppping -3 and 00", "_____no_output_____" ] ], [ [ "y_no_fail_ects = []\nx_no_fail = ['02','4','7','10','12']\n\n# Looking at each faculty \nfor i in tqdm_notebook(df['Fakultet'].unique()):\n df_faculty = df[df['Fakultet']==i]\n \n y_no_fail_ects_inner=[]\n \n for k in ['Weigthed_02','Weigthed_4','Weigthed_7','Weigthed_10','Weigthed_12']:\n y_no_fail_ects_inner.append(df_faculty[k].sum(skipna = True)) \n \n y_no_fail_ects.append(y_no_fail_ects_inner)\n \ny_no_fail_ects", "_____no_output_____" ], [ "# calc frequencies \ny_no_fail_ects_freq = []\n\n# running through each faculty\nfor i in range(len(y_ects)):\n \n y_no_fail_ects_freq_inner = []\n \n # calc frequencies \n for q in range(len(y_no_fail_ects[i])):\n y_no_fail_ects_freq_inner.append(y_no_fail_ects[i][q]/sum(y_no_fail_ects[i]))\n\n y_no_fail_ects_freq.append(y_no_fail_ects_freq_inner)\n\n\ny_no_fail_ects_freq", "_____no_output_____" ], [ "# This figure is used in the analysis\nf, ax = plt.subplots(figsize=(15,10))\n\nplt.subplot(2, 3, 1)\nplt.title(Faculty_names[0], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.35])\nplt.grid(axis ='y',zorder=0)\nplt.ylabel('Frequency',fontsize=14)\nplt.annotate('Number of passing grades: '+ str(list_no_fail_number_of_grades_faculties[0]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x_no_fail, y_no_fail_ects_freq[0], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 2)\nplt.title(Faculty_names[1], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.35])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of passing grades: '+ str(list_no_fail_number_of_grades_faculties[1]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x_no_fail, y_no_fail_ects_freq[1], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 3)\nplt.title(Faculty_names[2], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.35])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of passing grades: '+ str(list_no_fail_number_of_grades_faculties[2]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x_no_fail, y_no_fail_ects_freq[2], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 4)\nplt.title(Faculty_names[3], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.35])\nplt.grid(axis ='y',zorder=0)\nplt.ylabel('Frequency',fontsize=14)\nplt.annotate('Number of passing grades: '+ str(list_no_fail_number_of_grades_faculties[3]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x_no_fail, y_no_fail_ects_freq[3], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 5)\nplt.title(Faculty_names[4], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.35])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of passing grades: '+ str(list_no_fail_number_of_grades_faculties[4]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x_no_fail, y_no_fail_ects_freq[4], width=0.93, edgecolor='black',zorder=3)\n\nplt.subplot(2, 3, 6)\nplt.title(Faculty_names[5], fontsize = 16, weight = 'bold')\nplt.ylim([0,0.35])\nplt.grid(axis ='y',zorder=0)\nplt.annotate('Number of passing grades: '+ str(list_no_fail_number_of_grades_faculties[5]), (0,0), (0, -20),fontsize= 13, xycoords='axes fraction', textcoords='offset points', va='top')\nplt.bar(x_no_fail, y_no_fail_ects_freq[5], width=0.93, edgecolor='black',zorder=3)\n\nf.savefig('histogram_gades_split_faculty_ECTS_weight_NO_FAIL.png')", "_____no_output_____" ] ], [ [ "### Calculating GPA by faculty", "_____no_output_____" ] ], [ [ "from math import isnan\nimport math", "_____no_output_____" ], [ "#Calculate gpa when ONLY PASSED exams are counted\nsnit = []\nfor i in range(0,len(df)):\n x_02 = df[2][i]\n x_04 = df[4][i]\n x_07 = df[7][i]\n x_10 = df[10][i]\n x_12 = df[12][i]\n \n number = (x_12,x_10,x_07,x_04,x_02)\n grades = [12,10,7,4,2]\n mydick = dict(zip(grades,number))\n cleandick = {k: mydick[k] for k in mydick if not isnan(mydick[k])}\n \n num = sum([x * y for x,y in mydick.items()])\n den = sum(mydick.values())\n \n snit.append(num/den)\n \ndf[\"Snit\"] = snit", "_____no_output_____" ], [ "# Here I calculate the GPA of some form of assessment\ndef gpa(df,string):\n \n x_gpa = []\n x_sho = []\n x_ect = []\n\n for i in range(0,len(df)):\n if df[\"Fakultet\"][i] == string:\n if math.isnan(df[\"Snit\"][i]) == False:\n x_gpa.append(float(df[\"Snit\"][i]))\n x_sho.append(float(df[\"Fremmødte\"][i]))\n x_ect.append(float(df[\"Credit_edit\"][i]))\n\n den = 0\n num = 0\n for i in range(0,len(x_gpa)):\n den = x_sho[i]*x_ect[i] + den\n num = x_gpa[i]*x_sho[i]*x_ect[i] + num\n out = num/den\n return out", "_____no_output_____" ], [ "# Looping through each faculty\n\nfor i in df['Fakultet'].unique():\n print(gpa(df,i))", "_____no_output_____" ] ], [ [ "### Type of assessments broken down by faculties", "_____no_output_____" ] ], [ [ "list_type_ass=list(df['Type of assessmet_edit'].unique())", "_____no_output_____" ], [ "y_type_ass_inner = []\ny_type_ass = []\n\n# Looking at each faculty \nfor i in tqdm_notebook(df['Fakultet'].unique()):\n df_faculty = df[df['Fakultet']==i]\n \n # Running through each type of assessment.\n for k in list_type_ass:\n # Summing number of passed people for all courses (in each faculty) broken down on each tyep of assessment\n y_type_ass_inner.append(df_faculty[df_faculty['Type of assessmet_edit']==k]['Antal bestået'].sum(skipna = True)) \n \n y_type_ass.append(y_type_ass_inner)\n y_type_ass_inner=[]", "_____no_output_____" ], [ "# Creating the categories which we want for plot.\ncategories = []\n\nfor i in range(len(df['Fakultet'].unique())):\n \n categories_inner = []\n \n # Oral\n categories_inner.append(y_type_ass[i][1])\n \n # Written not under invigilation\n categories_inner.append(y_type_ass[i][2])\n \n # Written under invigilation\n categories_inner.append(y_type_ass[i][4])\n \n # Rest\n categories_inner.append(y_type_ass[i][0]+y_type_ass[i][3]+y_type_ass[i][5]+y_type_ass[i][6]+y_type_ass[i][7]\\\n +y_type_ass[i][8]+y_type_ass[i][9])\n \n categories.append(categories_inner)", "_____no_output_____" ], [ "#calc share.\n\nlist_categories_share = []\n\n# Running through each faculty\nfor i in range(len(categories)):\n \n categories_share_inner = []\n \n # For each faculty calc type of ass shares. \n for k in range(len(categories[i])):\n categories_share= categories[i][k]/sum(categories[i])*100 # times a 100 for %\n categories_share_inner.append(categories_share)\n \n list_categories_share.append(categories_share_inner) ", "_____no_output_____" ], [ "# Converting list to DataFrame \ndfcat= pd.DataFrame(list_categories_share) \ndfcat\ndfcat=dfcat.T\ndfcat.columns = ['Science','Social Sciences','Humanities','Health & Medical Sciences','Law','Theology']\ndfcat.rename(index={0:'Oral',1:'Written not under invigilation',2:'Written under invigilation',3:'Rest'})\n#dfcat.index.name = 'type_ass'\ndfcat=dfcat.T\ndfcat.columns = ['Oral','Written not invigilation','Written invigilation','Rest']", "_____no_output_____" ], [ "colors = [\"#011f4b\",\"#005b96\",\"#6497b1\",'#b3cde0']\n\n\ndfcat.plot(kind='bar', stacked=True, color = colors, fontsize = 12)\n#plt.legend(bbox_to_anchor=(0, 1), loc='upper left', ncol=1)\nplt.rcParams[\"figure.figsize\"] = [15,15]\nplt.legend(bbox_to_anchor=(0,1.02,1,0.2), loc=\"lower left\", mode=\"expand\", borderaxespad=0, ncol=2, fontsize = 12) #,weight = 'bold')\nplt.tight_layout()\nplt.savefig('stacked_bar_share_ass.png')", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/bhadreshpsavani/ExploringSentimentalAnalysis/blob/main/SentimentalAnalysisWithGPTNeo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "## Step1. Import and Load Data", "_____no_output_____" ] ], [ [ "!pip install -q pip install git+https://github.com/huggingface/transformers.git\n!pip install -q datasets", " Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n Preparing wheel metadata ... \u001b[?25l\u001b[?25hdone\n\u001b[K |████████████████████████████████| 901kB 6.5MB/s \n\u001b[K |████████████████████████████████| 3.3MB 16.3MB/s \n\u001b[?25h Building wheel for transformers (PEP 517) ... \u001b[?25l\u001b[?25hdone\n\u001b[K |████████████████████████████████| 235kB 5.2MB/s \n\u001b[K |████████████████████████████████| 112kB 8.9MB/s \n\u001b[K |████████████████████████████████| 245kB 9.9MB/s \n\u001b[?25h" ], [ "from datasets import load_dataset\nemotions = load_dataset(\"emotion\")", "_____no_output_____" ], [ "import torch\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")", "_____no_output_____" ] ], [ [ "## Step2. Preprocess Data", "_____no_output_____" ] ], [ [ "from transformers import AutoTokenizer\n\nmodel_name = \"EleutherAI/gpt-neo-125M\"\ntokenizer = AutoTokenizer.from_pretrained(model_name)\n\ndef tokenize(batch):\n return tokenizer(batch[\"text\"], padding=True, truncation=True)", "_____no_output_____" ], [ "tokenizer", "_____no_output_____" ], [ "tokenizer.add_special_tokens({'pad_token': '<|pad|>'})\ntokenizer", "_____no_output_____" ], [ "emotions_encoded = emotions.map(tokenize, batched=True, batch_size=None)", "_____no_output_____" ], [ "from transformers import AutoModelForSequenceClassification\nnum_labels = 6\nmodel = (AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=num_labels).to(device))", "_____no_output_____" ], [ "emotions_encoded[\"train\"].features", "_____no_output_____" ], [ "emotions_encoded.set_format(\"torch\", columns=[\"input_ids\", \"attention_mask\", \"label\"])\nemotions_encoded[\"train\"].features", "_____no_output_____" ], [ "from sklearn.metrics import accuracy_score, f1_score\n\ndef compute_metrics(pred):\n labels = pred.label_ids\n preds = pred.predictions.argmax(-1)\n f1 = f1_score(labels, preds, average=\"weighted\")\n acc = accuracy_score(labels, preds)\n return {\"accuracy\": acc, \"f1\": f1}", "_____no_output_____" ], [ "from transformers import Trainer, TrainingArguments\n\nbatch_size = 2\nlogging_steps = len(emotions_encoded[\"train\"]) // batch_size\ntraining_args = TrainingArguments(output_dir=\"results\",\n num_train_epochs=2,\n learning_rate=2e-5,\n per_device_train_batch_size=batch_size,\n per_device_eval_batch_size=batch_size,\n load_best_model_at_end=True,\n metric_for_best_model=\"f1\",\n weight_decay=0.01,\n evaluation_strategy=\"epoch\",\n disable_tqdm=False,\n logging_steps=logging_steps,)", "_____no_output_____" ], [ "from transformers import Trainer\n\ntrainer = Trainer(model=model, args=training_args,\n compute_metrics=compute_metrics,\n train_dataset=emotions_encoded[\"train\"],\n eval_dataset=emotions_encoded[\"validation\"])\ntrainer.train();", "_____no_output_____" ], [ "results = trainer.evaluate()\nresults", "_____no_output_____" ], [ "preds_output = trainer.predict(emotions_encoded[\"validation\"])\npreds_output.metrics", "_____no_output_____" ], [ "import numpy as np\nfrom sklearn.metrics import plot_confusion_matrix\ny_valid = np.array(emotions_encoded[\"validation\"][\"label\"])\ny_preds = np.argmax(preds_output.predictions, axis=1)\nlabels = ['sadness', 'joy', 'love', 'anger', 'fear', 'surprise']\nplot_confusion_matrix(y_preds, y_valid, labels)", "_____no_output_____" ], [ "model.save_pretrained('./model')\ntokenizer.save_pretrained('./model')", "_____no_output_____" ], [ "", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## parameters", "_____no_output_____" ] ], [ [ "CLUSTER_ALGO = 'KMedoids'\nC_SHAPE ='circle'\n#C_SHAPE ='CIRCLE'\n#C_SHAPE ='ellipse'\n#N_CLUSTERS = [50,300, 1000]\nN_CLUSTERS = [3]\nCLUSTERS_STD = 0.3\nN_P_CLUSTERS = [3, 30, 300, 3000]\nN_CLUSTERS_S = N_CLUSTERS[0]\nINNER_FOLDS = 3\nOUTER_FOLDS = 3", "_____no_output_____" ] ], [ [ "## includes", "_____no_output_____" ] ], [ [ "%matplotlib inline\n\nimport matplotlib\n\nimport matplotlib.pyplot as plt\nfrom sklearn.datasets import load_breast_cancer\nfrom sklearn.datasets import load_iris\nfrom sklearn.datasets import make_blobs\nfrom sklearn.datasets import make_moons\n", "_____no_output_____" ], [ "%load_ext autoreload\n%autoreload 2\npackages = !conda list\npackages", "_____no_output_____" ] ], [ [ "## Output registry", "_____no_output_____" ] ], [ [ "from __future__ import print_function\nimport sys, os\n\nold__file__ = !pwd\n__file__ = !cd ../../../photon ;pwd\n#__file__ = !pwd\n__file__ = __file__[0]\n__file__\nsys.path.append(__file__)\nprint(sys.path)\nos.chdir(old__file__[0])\n!pwd\nold__file__[0]", "['/docker/photon_experiments/Cluster/kmeans-kmedoids', '/opt/conda/lib/python37.zip', '/opt/conda/lib/python3.7', '/opt/conda/lib/python3.7/lib-dynload', '', '/opt/conda/lib/python3.7/site-packages', '/opt/conda/lib/python3.7/site-packages/IPython/extensions', '/home/jovyan/.ipython', '/docker/photon']\n/docker/photon_experiments/Cluster/kmeans-kmedoids\n" ], [ "\nimport seaborn as sns; sns.set() # for plot styling\nimport numpy as np\nimport pandas as pd\nfrom math import floor,ceil\nfrom sklearn.model_selection import KFold\nfrom sklearn.manifold import TSNE\nimport itertools\nimport seaborn as sns\nimport matplotlib.pyplot as plt\n%matplotlib inline\n\n#set font size of labels on matplotlib plots\nplt.rc('font', size=16)\n\n#set style of plots\nsns.set_style('white')\n\n#define a custom palette\nPALLET = ['#40111D', '#DCD5E4', '#E7CC74'\n ,'#39C8C6', '#AC5583', '#D3500C'\n ,'#FFB139', '#98ADA7', '#AD989E'\n ,'#708090','#6C8570','#3E534D'\n ,'#0B8FD3','#0B47D3','#96D30B' \n ,'#630C3A','#F1D0AF','#64788B' \n ,'#8B7764','#7A3C5D','#77648B'\n ,'#eaff39','#39ff4e','#4e39ff'\n ,'#ff4e39','#87ff39','#ff3987', ]\nN_PALLET = len(PALLET)\nsns.set_palette(PALLET)\nsns.palplot(PALLET)\n\n\nfrom clusim.clustering import Clustering, remap2match\nimport clusim.sim as sim\n\nfrom photonai.base import Hyperpipe, PipelineElement, Preprocessing, OutputSettings\nfrom photonai.optimization import FloatRange, Categorical, IntegerRange\nfrom photonai.base.photon_elements import PhotonRegistry\nfrom photonai.visual.graphics import plot_cm\nfrom photonai.photonlogger.logger import logger\n#from photonai.base.registry.registry import PhotonRegistry", "/opt/conda/lib/python3.7/site-packages/sklearn/externals/joblib/__init__.py:15: DeprecationWarning: sklearn.externals.joblib is deprecated in 0.21 and will be removed in 0.23. Please import this functionality directly from joblib, which can be installed with: pip install joblib. If this warning is raised when loading pickled models, you may need to re-serialize those models with scikit-learn 0.21+.\n warnings.warn(msg, category=DeprecationWarning)\n" ] ], [ [ "## function defintions", "_____no_output_____" ] ], [ [ "def yield_parameters_ellipse(n_p_clusters):\n cluster_std = CLUSTERS_STD\n for n_p_cluster in n_p_clusters:\n for n_cluster in N_CLUSTERS:\n print('ncluster:', n_cluster)\n n_cluster_std = [cluster_std for k in range(n_cluster)]\n n_samples = [n_p_cluster for k in range(n_cluster)]\n data_X, data_y = make_blobs(n_samples=n_samples,\n cluster_std=n_cluster_std, random_state=0)\n transformation = [[0.6, -0.6], [-0.4, 0.8]]\n X_ellipse = np.dot(data_X, transformation)\n yield [X_ellipse, data_y, n_cluster]", "_____no_output_____" ], [ "def yield_parameters(n_p_clusters):\n cluster_std = CLUSTERS_STD\n for n_p_cluster in n_p_clusters:\n for n_cluster in N_CLUSTERS:\n n_cluster_std = [cluster_std for k in range(n_cluster)]\n n_samples = [n_p_cluster for k in range(n_cluster)]\n data_X, data_y = make_blobs(n_samples=n_samples,\n cluster_std=n_cluster_std, random_state=0)\n yield [data_X, data_y, n_cluster]", "_____no_output_____" ], [ "def results_to_df(results):\n ll = []\n for obj in results:\n ll.append([obj.operation,\n obj.value,\n obj.metric_name])\n _results=pd.DataFrame(ll).pivot(index=2, columns=0, values=1)\n _results.columns=['Mean','STD']\n \n return(_results)", "_____no_output_____" ], [ "def cluster_plot(my_pipe, data_X, n_cluster, PALLET):\n y_pred= my_pipe.predict(data_X)\n data = pd.DataFrame(data_X[:, 0],columns=['x'])\n data['y'] = data_X[:, 1]\n data['labels'] = y_pred\n facet = sns.lmplot(data=data, x='x', y='y', hue='labels', \n aspect= 1.0, height=10,\n fit_reg=False, legend=True, legend_out=True)\n\n customPalette = PALLET #*ceil((n_cluster/N_PALLET +2))\n for i, label in enumerate( np.sort(data['labels'].unique())):\n plt.annotate(label, \n data.loc[data['labels']==label,['x','y']].mean(),\n horizontalalignment='center',\n verticalalignment='center',\n size=5, weight='bold',\n color='white',\n backgroundcolor=customPalette[i]) \n\n plt.show()\n return y_pred", "_____no_output_____" ], [ "def simple_output(string: str, number: int) -> None:\n print(string, number)\n logger.info(string, number )", "_____no_output_____" ], [ "__file__ = \"exp1.log\"\nbase_folder = os.path.dirname(os.path.abspath(''))\ncustom_elements_folder = os.path.join(base_folder, 'custom_elements')\ncustom_elements_folder", "_____no_output_____" ], [ "registry = PhotonRegistry(custom_elements_folder=custom_elements_folder)\nregistry.activate()\nregistry.PHOTON_REGISTRIES,PhotonRegistry.PHOTON_REGISTRIES", "_____no_output_____" ], [ "registry.activate()\nregistry.list_available_elements()\n# take off last name", "\nPhotonCore\nARDRegression sklearn.linear_model.ARDRegression Estimator\nAdaBoostClassifier sklearn.ensemble.AdaBoostClassifier Estimator\nAdaBoostRegressor sklearn.ensemble.AdaBoostRegressor Estimator\nBaggingClassifier sklearn.ensemble.BaggingClassifier Estimator\nBaggingRegressor sklearn.ensemble.BaggingRegressor Estimator\nBayesianGaussianMixture sklearn.mixture.BayesianGaussianMixture Estimator\nBayesianRidge sklearn.linear_model.BayesianRidge Estimator\nBernoulliNB sklearn.naive_bayes.BernoulliNB Estimator\nBernoulliRBM sklearn.neural_network.BernoulliRBM Estimator\nBinarizer sklearn.preprocessing.Binarizer Transformer\nCCA sklearn.cross_decomposition.CCA Transformer\nConfounderRemoval photonai.modelwrapper.ConfounderRemoval.ConfounderRemoval Transformer\nDecisionTreeClassifier sklearn.tree.DecisionTreeClassifier Estimator\nDecisionTreeRegressor sklearn.tree.DecisionTreeRegressor Estimator\nDictionaryLearning sklearn.decomposition.DictionaryLearning Transformer\nDummyClassifier sklearn.dummy.DummyClassifier Estimator\nDummyRegressor sklearn.dummy.DummyRegressor Estimator\nElasticNet sklearn.linear_model.ElasticNet Estimator\nExtraDecisionTreeClassifier sklearn.tree.ExtraDecisionTreeClassifier Estimator\nExtraDecisionTreeRegressor sklearn.tree.ExtraDecisionTreeRegressor Estimator\nExtraTreesClassifier sklearn.ensemble.ExtraTreesClassifier Estimator\nExtraTreesRegressor sklearn.ensemble.ExtraTreesRegressor Estimator\nFClassifSelectPercentile photonai.modelwrapper.FeatureSelection.FClassifSelectPercentile Transformer\nFRegressionFilterPValue photonai.modelwrapper.FeatureSelection.FRegressionFilterPValue Transformer\nFRegressionSelectPercentile photonai.modelwrapper.FeatureSelection.FRegressionSelectPercentile Transformer\nFactorAnalysis sklearn.decomposition.FactorAnalysis Transformer\nFastICA sklearn.decomposition.FastICA Transformer\nFeatureEncoder photonai.modelwrapper.OrdinalEncoder.FeatureEncoder Transformer\nFunctionTransformer sklearn.preprocessing.FunctionTransformer Transformer\nGaussianMixture sklearn.mixture.GaussianMixture Estimator\nGaussianNB sklearn.naive_bayes.GaussianNB Estimator\nGaussianProcessClassifier sklearn.gaussian_process.GaussianProcessClassifier Estimator\nGaussianProcessRegressor sklearn.gaussian_process.GaussianProcessRegressor Estimator\nGenericUnivariateSelect sklearn.feature_selection.GenericUnivariateSelect Transformer\nGradientBoostingClassifier sklearn.ensemble.GradientBoostingClassifier Estimator\nGradientBoostingRegressor sklearn.ensemble.GradientBoostingRegressor Estimator\nHuberRegressor sklearn.linear_model.HuberRegressor Estimator\nImbalancedDataTransformer photonai.modelwrapper.imbalanced_data_transformer.ImbalancedDataTransformer Transformer\nIncrementalPCA sklearn.decomposition.IncrementalPCA Transformer\nKNeighborsClassifier sklearn.neighbors.KNeighborsClassifier Estimator\nKNeighborsRegressor sklearn.neighbors.KNeighborsRegressor Estimator\nKerasBaseClassifier photonai.modelwrapper.keras_base_models.KerasBaseClassifier Estimator\nKerasBaseRegression photonai.modelwrapper.keras_base_models.KerasBaseRegression Estimator\nKerasDnnClassifier photonai.modelwrapper.keras_dnn_classifier.KerasDnnClassifier Estimator\nKerasDnnRegressor photonai.modelwrapper.keras_dnn_regressor.KerasDnnRegressor Estimator\nKernelCenterer sklearn.preprocessing.KernelCenterer Transformer\nKernelPCA sklearn.decomposition.KernelPCA Transformer\nKernelRidge sklearn.kernel_ridge.KernelRidge Estimator\nLabelEncoder photonai.modelwrapper.LabelEncoder.LabelEncoder Transformer\nLars sklearn.linear_model.Lars Estimator\nLasso sklearn.linear_model.Lasso Estimator\nLassoFeatureSelection photonai.modelwrapper.FeatureSelection.LassoFeatureSelection Transformer\nLassoLars sklearn.linear_model.LassoLars Estimator\nLatentDirichletAllocation sklearn.decomposition.LatentDirichletAllocation Transformer\nLinearRegression sklearn.linear_model.LinearRegression Estimator\nLinearSVC sklearn.svm.LinearSVC Estimator\nLinearSVR sklearn.svm.LinearSVR Estimator\nLogisticRegression sklearn.linear_model.LogisticRegression Estimator\nMLPClassifier sklearn.neural_network.MLPClassifier Estimator\nMLPRegressor sklearn.neural_network.MLPRegressor Estimator\nMaxAbsScaler sklearn.preprocessing.MaxAbsScaler Transformer\nMinMaxScaler sklearn.preprocessing.MinMaxScaler Transformer\nMiniBatchDictionaryLearning sklearn.decomposition.MiniBatchDictionaryLearning Transformer\nMiniBatchSparsePCA sklearn.decomposition.MiniBatchSparsePCA Transformer\nMultinomialNB sklearn.naive_bayes.MultinomialNB Estimator\nNMF sklearn.decompositcion.NMF Transformer\nNearestCentroid sklearn.neighbors.NearestCentroid Estimator\nNormalizer sklearn.preprocessing.Normalizer Transformer\nNuSVC sklearn.svm.NuSVC Estimator\nNuSVR sklearn.svm.NuSVR Estimator\nOneClassSVM sklearn.svm.OneClassSVM Estimator\nPCA sklearn.decomposition.PCA Transformer\nPLSCanonical sklearn.cross_decomposition.PLSCanonical Transformer\nPLSRegression sklearn.cross_decomposition.PLSRegression Transformer\nPLSSVD sklearn.cross_decomposition.PLSSVD Transformer\nPassiveAggressiveClassifier sklearn.linear_model.PassiveAggressiveClassifier Estimator\nPassiveAggressiveRegressor sklearn.linear_model.PassiveAggressiveRegressor Estimator\nPerceptron sklearn.linear_model.Perceptron Estimator\nPhotonMLPClassifier photonai.modelwrapper.PhotonMLPClassifier.PhotonMLPClassifier Estimator\nPhotonOneClassSVM photonai.modelwrapper.PhotonOneClassSVM.PhotonOneClassSVM Estimator\nPhotonTestXPredictor photonai.test.processing_tests.results_tests.XPredictor Estimator\nPhotonVotingClassifier photonai.modelwrapper.Voting.PhotonVotingClassifier Estimator\nPhotonVotingRegressor photonai.modelwrapper.Voting.PhotonVotingRegressor Estimator\nPolynomialFeatures sklearn.preprocessing.PolynomialFeatures Transformer\nPowerTransformer sklearn.preprocessing.PowerTransformer Transformer\nQuantileTransformer sklearn.preprocessing.QuantileTransformer Transformer\nRANSACRegressor sklearn.linear_model.RANSACRegressor Estimator\nRFE sklearn.feature_selection.RFE Transformer\nRFECV sklearn.feature_selection.RFECV Transformer\nRadiusNeighborsClassifier sklearn.neighbors.RadiusNeighborsClassifier Estimator\nRadiusNeighborsRegressor sklearn.neighbors.RadiusNeighborsRegressor Estimator\nRandomForestClassifier sklearn.ensemble.RandomForestClassifier Estimator\nRandomForestRegressor sklearn.ensemble.RandomForestRegressor Estimator\nRandomTreesEmbedding sklearn.ensemble.RandomTreesEmbedding Transformer\nRangeRestrictor photonai.modelwrapper.RangeRestrictor.RangeRestrictor Estimator\nRidge sklearn.linear_model.Ridge Estimator\nRidgeClassifier sklearn.linear_model.RidgeClassifier Estimator\nRobustScaler sklearn.preprocessing.RobustScaler Transformer\nSGDClassifier sklearn.linear_model.SGDClassifier Estimator\nSGDRegressor sklearn.linear_model.SGDRegressor Estimator\nSVC sklearn.svm.SVC Estimator\nSVR sklearn.svm.SVR Estimator\nSamplePairingClassification photonai.modelwrapper.SamplePairing.SamplePairingClassification Transformer\nSamplePairingRegression photonai.modelwrapper.SamplePairing.SamplePairingRegression Transformer\nSelectFdr sklearn.feature_selection.SelectFdr Transformer\nSelectFpr sklearn.feature_selection.SelectFpr Transformer\nSelectFromModel sklearn.feature_selection.SelectFromModel Transformer\nSelectFwe sklearn.feature_selection.SelectFwe Transformer\nSelectKBest sklearn.feature_selection.SelectKBest Transformer\nSelectPercentile sklearn.feature_selection.SelectPercentile Transformer\nSimpleImputer sklearn.impute.SimpleImputer Transformer\nSourceSplitter photonai.modelwrapper.source_splitter.SourceSplitter Transformer\nSparseCoder sklearn.decomposition.SparseCoder Transformer\nSparsePCA sklearn.decomposition.SparsePCA Transformer\nStandardScaler sklearn.preprocessing.StandardScaler Transformer\nTheilSenRegressor sklearn.linear_model.TheilSenRegressor Estimator\nTruncatedSVD sklearn.decomposition.TruncatedSVD Transformer\nVarianceThreshold sklearn.feature_selection.VarianceThreshold Transformer\ndict_learning sklearn.decomposition.dict_learning Transformer\ndict_learning_online sklearn.decomposition.dict_learning_online Transformer\nfastica sklearn.decomposition.fastica Transformer\nsparse_encode sklearn.decomposition.sparse_encode Transformer\n\nPhotonCluster\nKMeans sklearn.cluster.KMeans Estimator\nKMedoids sklearn_extra.cluster.KMedoids Estimator\n\nPhotonNeuro\nBrainAtlas photonai.neuro.brain_atlas.BrainAtlas Transformer\nBrainMask photonai.neuro.brain_atlas.BrainMask Transformer\nPatchImages photonai.neuro.nifti_transformations.PatchImages Transformer\nResampleImages photonai.neuro.nifti_transformations.ResampleImages Transformer\nSmoothImages photonai.neuro.nifti_transformations.SmoothImages Transformer\n" ] ], [ [ "## KMeans blobs", "_____no_output_____" ] ], [ [ "\nregistry.info(CLUSTER_ALGO)", "----------------------------------\nName: KMedoids\nNamespace: sklearn_extra.cluster\n----------------------------------\nPossible Hyperparameters as derived from constructor:\nn_clusters n_clusters=8 \nmetric metric='euclidean' \ninit init='heuristic' \nmax_iter max_iter=300 \nrandom_state random_state=None \n----------------------------------\n" ], [ "def hyper_cluster(cluster_name):\n if C_SHAPE == 'ellipse' :\n yield_cluster = yield_parameters_ellipse\n else: \n yield_cluster = yield_parameters\n \n n_p_clusters = N_P_CLUSTERS\n for data_X, data_y,n_cluster in yield_cluster(n_p_clusters):\n simple_output('CLUSTER_ALGO:', CLUSTER_ALGO)\n simple_output('C_SHAPE:',C_SHAPE)\n simple_output('n_cluster:', n_cluster)\n simple_output('CLUSTERS_STD:', CLUSTERS_STD)\n \n simple_output('INNER_FOLDS:', INNER_FOLDS)\n simple_output('OUTER_FOLDS:', OUTER_FOLDS) \n simple_output('n_points:', len(data_y))\n\n X = data_X.copy(); y = data_y.copy()\n # DESIGN YOUR PIPELINE\n settings = OutputSettings(project_folder='./tmp/')\n \n my_pipe = Hyperpipe('batching',\n optimizer='sk_opt',\n # optimizer_params={'n_configurations': 25},\n metrics=['ARI', 'MI', 'HCV', 'FM'],\n best_config_metric='ARI',\n outer_cv=KFold(n_splits=OUTER_FOLDS),\n inner_cv=KFold(n_splits=INNER_FOLDS),\n verbosity=0,\n output_settings=settings)\n\n\n my_pipe += PipelineElement(cluster_name, hyperparameters={\n 'n_clusters': IntegerRange(floor(n_cluster*.7)\n , ceil(n_cluster*1.2)),\n },random_state=777)\n\n logger.info('Cluster optimization range:', floor(n_cluster*.7), ceil(n_cluster*1.2))\n print('Cluster optimization range:', floor(n_cluster*.7), ceil(n_cluster*1.2)) \n\n # NOW TRAIN YOUR PIPELINE\n my_pipe.fit(X, y)\n\n debug = True\n #------------------------------plot\n y_pred=cluster_plot(my_pipe, X, n_cluster, PALLET)\n #--------------------------------- best\n print(pd.DataFrame(my_pipe.best_config.items()\n ,columns=['n_clusters', 'k']))\n #------------------------------\n print('train','\\n'\n ,results_to_df(my_pipe.results.metrics_train))\n print('test','\\n'\n ,results_to_df(my_pipe.results.metrics_test))\n #------------------------------ \n # turn the ground-truth labels into a clusim Clustering\n true_clustering = Clustering().from_membership_list(y) \n kmeans_clustering = Clustering().from_membership_list(y_pred) # lets see how similar the predicted k-means clustering is to the true clustering\n #------------------------------\n # using all available similar measures!\n row_format2 =\"{:>25}\" * (2)\n for simfunc in sim.available_similarity_measures:\n print(row_format2.format(simfunc, eval('sim.' + simfunc+'(true_clustering, kmeans_clustering)')))\n #------------------------------# The element-centric similarity is particularly useful for understanding\n # how a clustering method performed\n\n # Let's start with the single similarity value:\n elsim = sim.element_sim(true_clustering, kmeans_clustering)\n print(\"Element-centric similarity: {}\".format(elsim))", "_____no_output_____" ], [ "hyper_cluster(CLUSTER_ALGO)", "CLUSTER_ALGO: KMedoids\nC_SHAPE: circle\nn_cluster: 3\nINNER_FOLDS: 3\nOUTER_FOLDS: 3\nn_points: 9\nCluster optimization range: 2 4\n***************************************************************************************************************\nPHOTON ANALYSIS: batching\n***************************************************************************************************************\n\n********************************************************\nOuter Cross validation Fold 1\n********************************************************\n---------------------------------------------------------------------------------------------------------------\nBEST_CONFIG \n---------------------------------------------------------------------------------------------------------------\n{\n \"KMedoids\": [\n \"n_clusters=4\"\n ]\n}\n+--------+-------------------+------------------+\n| METRIC | PERFORMANCE TRAIN | PERFORMANCE TEST |\n+--------+-------------------+------------------+\n| ARI | 0.7619 | 1.0000 |\n| MI | 0.6154 | 1.0000 |\n| HCV | 1.0000 | 1.0000 |\n| FM | 0.8165 | 0.0000 |\n+--------+-------------------+------------------+\n\n********************************************************\nOuter Cross validation Fold 2\n********************************************************\n---------------------------------------------------------------------------------------------------------------\nBEST_CONFIG \n---------------------------------------------------------------------------------------------------------------\n{\n \"KMedoids\": [\n \"n_clusters=4\"\n ]\n}\n+--------+-------------------+------------------+\n| METRIC | PERFORMANCE TRAIN | PERFORMANCE TEST |\n+--------+-------------------+------------------+\n| ARI | 0.7619 | 1.0000 |\n| MI | 0.6154 | 1.0000 |\n| HCV | 1.0000 | 1.0000 |\n| FM | 0.8165 | 0.0000 |\n+--------+-------------------+------------------+\n\n********************************************************\nOuter Cross validation Fold 3\n********************************************************\n---------------------------------------------------------------------------------------------------------------\nBEST_CONFIG \n---------------------------------------------------------------------------------------------------------------\n{\n \"KMedoids\": [\n \"n_clusters=4\"\n ]\n}\n+--------+-------------------+------------------+\n| METRIC | PERFORMANCE TRAIN | PERFORMANCE TEST |\n+--------+-------------------+------------------+\n| ARI | 0.7619 | 1.0000 |\n| MI | 0.6154 | 1.0000 |\n| HCV | 1.0000 | 1.0000 |\n| FM | 0.8165 | 0.0000 |\n+--------+-------------------+------------------+\n\n===============================================================================================================\nOVERALL BEST CONFIGURATION\n===============================================================================================================\n{\n \"KMedoids\": [\n \"n_clusters=4\"\n ]\n}\n\nAnalysis batching done in 0:00:24.506777\nYour results are stored in ./tmp/batching_results_2020-06-04_16-20-14\n***************************************************************************************************************\nPHOTON 1.0.0b - www.photon-ai.com \n n_clusters k\n0 KMedoids__n_clusters 4\ntrain \n Mean STD\n2 \nARI 0.761905 0.000000e+00\nFM 0.816497 1.110223e-16\nHCV 1.000000 0.000000e+00\nMI 0.615385 0.000000e+00\ntest \n Mean STD\n2 \nARI 1.0 0.0\nFM 0.0 0.0\nHCV 1.0 0.0\nMI 1.0 0.0\n jaccard_index 0.7777777777777778\n rand_index 0.9444444444444444\n adjrand_index 0.8399999999999999\n fowlkes_mallows_index 0.8819171036881969\n fmeasure 0.875\n purity_index 0.8888888888888888\n classification_error 0.11111111111111116\n czekanowski_index 0.875\n dice_index 0.875\n sorensen_index 0.875\n rogers_tanimoto_index 0.8947368421052632\n southwood_index 3.5\n pearson_correlation 0.0038314176245210726\n corrected_chance 0.7446541086468912\n sample_expected_sim 0.0\n nmi 0.9119400080526283\n mi 1.5849625007211563\n adj_mi 0.8546410755834939\n rmi 0.4545024595933558\n vi 0.30609861135149674\n geometric_accuracy 0.9428090415820634\n overlap_quality -0.0\n onmi 0.8869847631644618\n omega_index 0.84\nElement-centric similarity: 0.8518518518518519\nCLUSTER_ALGO: KMedoids\nC_SHAPE: circle\nn_cluster: 3\nINNER_FOLDS: 3\nOUTER_FOLDS: 3\nn_points: 90\nCluster optimization range: 2 4\n***************************************************************************************************************\nPHOTON ANALYSIS: batching\n***************************************************************************************************************\n\n********************************************************\nOuter Cross validation Fold 1\n********************************************************\n---------------------------------------------------------------------------------------------------------------\nBEST_CONFIG \n---------------------------------------------------------------------------------------------------------------\n{\n \"KMedoids\": [\n \"n_clusters=3\"\n ]\n}\n+--------+-------------------+------------------+\n| METRIC | PERFORMANCE TRAIN | PERFORMANCE TEST |\n+--------+-------------------+------------------+\n| ARI | 1.0000 | 1.0000 |\n| MI | 1.0000 | 1.0000 |\n| HCV | 1.0000 | 1.0000 |\n| FM | 1.0000 | 1.0000 |\n+--------+-------------------+------------------+\n\n********************************************************\nOuter Cross validation Fold 2\n********************************************************\n---------------------------------------------------------------------------------------------------------------\nBEST_CONFIG \n---------------------------------------------------------------------------------------------------------------\n{\n \"KMedoids\": [\n \"n_clusters=2\"\n ]\n}\n+--------+-------------------+------------------+\n| METRIC | PERFORMANCE TRAIN | PERFORMANCE TEST |\n+--------+-------------------+------------------+\n| ARI | 0.6397 | 0.3800 |\n| MI | 0.5713 | 0.4221 |\n| HCV | 0.5782 | 0.4455 |\n| FM | 0.8002 | 0.7184 |\n+--------+-------------------+------------------+\n\n********************************************************\nOuter Cross validation Fold 3\n********************************************************\n---------------------------------------------------------------------------------------------------------------\nBEST_CONFIG \n---------------------------------------------------------------------------------------------------------------\n{\n \"KMedoids\": [\n \"n_clusters=3\"\n ]\n}\n+--------+-------------------+------------------+\n| METRIC | PERFORMANCE TRAIN | PERFORMANCE TEST |\n+--------+-------------------+------------------+\n| ARI | 1.0000 | 1.0000 |\n| MI | 1.0000 | 1.0000 |\n| HCV | 1.0000 | 1.0000 |\n| FM | 1.0000 | 1.0000 |\n+--------+-------------------+------------------+\n\n===============================================================================================================\nOVERALL BEST CONFIGURATION\n===============================================================================================================\n{\n \"KMedoids\": [\n \"n_clusters=3\"\n ]\n}\n" ], [ "Cottman\n*-+\n*+-=END", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "from collections import OrderedDict\nfrom collections import namedtuple\n\nimport numpy as np\nfrom scipy import stats\n\n\n# R precision\ndef r_precision(targets, predictions, max_n_predictions=500):\n # Assumes predictions are sorted by relevance\n # First, cap the number of predictions\n predictions = predictions[:max_n_predictions]\n\n # Calculate metric\n target_set = set(targets)\n target_count = len(target_set)\n return float(len(set(predictions[:target_count]).intersection(target_set))) / target_count\n\ndef dcg(relevant_elements, retrieved_elements, k, *args, **kwargs):\n \"\"\"Compute the Discounted Cumulative Gain.\n Rewards elements being retrieved in descending order of relevance.\n \\[ DCG = rel_1 + \\sum_{i=2}^{|R|} \\frac{rel_i}{\\log_2(i + 1)} \\]\n Args:\n retrieved_elements (list): List of retrieved elements\n relevant_elements (list): List of relevant elements\n k (int): 1-based index of the maximum element in retrieved_elements\n taken in the computation\n Note: The vector `retrieved_elements` is truncated at first, THEN\n deduplication is done, keeping only the first occurence of each element.\n Returns:\n DCG value\n \"\"\"\n retrieved_elements = __get_unique(retrieved_elements[:k])\n relevant_elements = __get_unique(relevant_elements)\n if len(retrieved_elements) == 0 or len(relevant_elements) == 0:\n return 0.0\n # Computes an ordered vector of 1.0 and 0.0\n score = [float(el in relevant_elements) for el in retrieved_elements]\n # return score[0] + np.sum(score[1:] / np.log2(\n # 1 + np.arange(2, len(score) + 1)))\n return np.sum(score / np.log2(1 + np.arange(1, len(score) + 1)))\n\n\ndef ndcg(relevant_elements, retrieved_elements, k, *args, **kwargs):\n \"\"\"Compute the Normalized Discounted Cumulative Gain.\n Rewards elements being retrieved in descending order of relevance.\n The metric is determined by calculating the DCG and dividing it by the\n ideal or optimal DCG in the case that all recommended tracks are relevant.\n Note:\n The ideal DCG or IDCG is on our case equal to:\n \\[ IDCG = 1+\\sum_{i=2}^{min(\\left| G \\right|, k)}\\frac{1}{\\log_2(i +1)}\\]\n If the size of the set intersection of \\( G \\) and \\( R \\), is empty, then\n the IDCG is equal to 0. The NDCG metric is now calculated as:\n \\[ NDCG = \\frac{DCG}{IDCG + \\delta} \\]\n with \\( \\delta \\) a (very) small constant.\n The vector `retrieved_elements` is truncated at first, THEN\n deduplication is done, keeping only the first occurence of each element.\n Args:\n retrieved_elements (list): List of retrieved elements\n relevant_elements (list): List of relevant elements\n k (int): 1-based index of the maximum element in retrieved_elements\n taken in the computation\n Returns:\n NDCG value\n \"\"\"\n\n # TODO: When https://github.com/scikit-learn/scikit-learn/pull/9951 is\n # merged...\n idcg = dcg(\n relevant_elements, relevant_elements, min(k, len(relevant_elements)))\n if idcg == 0:\n raise ValueError(\"relevent_elements is empty, the metric is\"\n \"not defined\")\n true_dcg = dcg(relevant_elements, retrieved_elements, k)\n return true_dcg / idcg\n\n\ndef __get_unique(original_list):\n \"\"\"Get only unique values of a list but keep the order of the first\n occurence of each element\n \"\"\"\n return list(OrderedDict.fromkeys(original_list))\n\n\nMetrics = namedtuple('Metrics', ['r_precision', 'ndcg', 'plex_clicks'])\n\n\n# playlist extender clicks\ndef playlist_extender_clicks(targets, predictions, max_n_predictions=500):\n # Assumes predictions are sorted by relevance\n # First, cap the number of predictions\n predictions = predictions[:max_n_predictions]\n\n # Calculate metric\n i = set(predictions).intersection(set(targets))\n for index, t in enumerate(predictions):\n for track in i:\n if t == track:\n return float(int(index / 10))\n return float(max_n_predictions / 10.0 + 1)\n\n\n# def compute all metrics\ndef get_all_metrics(targets, predictions, k):\n return Metrics(r_precision(targets, predictions, k),\n ndcg(targets, predictions, k),\n playlist_extender_clicks(targets, predictions, k))\n\n\nMetricsSummary = namedtuple('MetricsSummary', ['mean_r_precision',\n 'mean_ndcg',\n 'mean_plex_clicks',\n 'coverage'])\n", "_____no_output_____" ], [ "#java -jar RankLib-2.10.jar -train BigRecall-TrainingFile750-2080.txt -ranker 6 -metric2t NDCG@20 -save BigRecallTrain750-2080Model-1Trees-NDCG20-tc1-lr05-leaf5.txt -tree 1 -tc 1 -shrinkage 0.5 -leaf 5\n#java -jar RankLib-2.10.jar -load BigRecallTrain750-2080Model-500Trees-NDCG20-tc1-lr05-leaf50.txt -rank BigRecallTestingFile750.txt -score BRScores750-2080Model-500Trees-NDCG20-tc1-lr05-leaf50.txt\n\n#java -jar RankLib-2.10.jar -train BigRecall-TrainingFile750-2080.txt -ranker 6 -metric2t NDCG@20 -save BigRecallTrain750-2080Model-500Trees-NDCG20-tc1-lr05-leaf50.txt -tree 500 -tc 1 -shrinkage 0.5 -leaf 50\n\n\nimport os\nTrainingFile='./Training/BigRecall-TrainingFile750-2080.txt'\nTestingFile='./Training/BigRecallTestingFile750.txt'\ntrees=[500]\ntcVals=[-1]\nshrinkages=[0.5]\nleaves= [50]\n\n\ndef createCommand(tree,tc,lr,leaf):\n opModelFile= './Training/ModelsAndScores/BigRecallTrain750-2080Model-'+str(tree)+'Trees-NDCG20-tc'+str(tc)+'-lr'+str(lr).replace('.','')+'-leaf'+str(leaf)+'.txt'\n trainCommand= 'java -jar ./Training/RankLib-2.10.jar -train ./Training/BigRecall-TrainingFile750-2080.txt -ranker 6 -metric2t NDCG@20 -silent -save '+ opModelFile+ ' -tree '+str(tree)+' -tc '+str(tc)+ ' -shrinkage '+str(lr)+ ' -leaf '+ str(leaf) +' -missingzero'\n #BRScores750-2080Model-1Trees-NDCG20-tc1-lr05-leaf5.txt\n opScoresFile='./Training/ModelsAndScores/BRScores750-2080Model-'+opModelFile.split('Model-')[1]\n testCommand= 'java -jar ./Training/RankLib-2.10.jar -load '+opModelFile+' -rank ./Training/BigRecallTestingFile750.txt -score '+opScoresFile\n return (opModelFile,trainCommand, opScoresFile, testCommand)\nparamSweep=[]\nfor tree in trees:\n for tc in tcVals:\n for lr in shrinkages:\n for leaf in leaves:\n paramSweep.append(createCommand(tree,tc,lr,leaf))", "_____no_output_____" ], [ "import multiprocessing as mp\nimport codecs\nimport os\nimport subprocess\n\ndef ExecuteRanklib(execTuples):\n try:\n trainCommand=execTuples[1]\n train= subprocess.check_output(trainCommand.split())\n print train\n print '----------'\n scoreCommand=execTuples[3]\n test=subprocess.check_output(scoreCommand.split())\n print test\n print '----------'\n \n except:\n print execTuples\n\n\npool = mp.Pool(processes=10)\npool.map(ExecuteRanklib, paramSweep)", "Process PoolWorker-30:\nProcess PoolWorker-23:\nProcess PoolWorker-24:\nProcess PoolWorker-27:\nProcess PoolWorker-26:\nProcess PoolWorker-25:\nProcess PoolWorker-28:\n" ], [ "TestFile='./Training/BigRecallTestingFile750.txt'\n\nwith open(TestFile) as f:\n test = f.readlines()\n\nPidTestTracks={}\nfor l in test:\n pid=l.split()[1].split(':')[1].strip()\n track=l.split('#')[1].strip()\n PidTestTracks.setdefault(pid,[]).append(track)\n \n###skip\nimport os\n\nMeta1Resultspath='/home/ubuntu/SpotifyChallenge/notebooks/Reranking/TestingQueryResults/Meta1/'\nMeta2Resultspath='/home/ubuntu/SpotifyChallenge/notebooks/Reranking/TestingQueryResults/Meta2/'\nQEPRFResultspath='/home/ubuntu/SpotifyChallenge/notebooks/Reranking/TestingQueryResults/QEPRF750/'\n\nMeta1Files=[Meta1Resultspath+x for x in os.listdir(Meta1Resultspath)]\nMeta2Files=[Meta2Resultspath+x for x in os.listdir(Meta2Resultspath)]\nQEPRFFiles=[QEPRFResultspath+x for x in os.listdir(QEPRFResultspath)]\n\n###skip\nimport codecs\ndef parseMetaFiles(path):\n playlistId=path.split('/')[-1].split('.op')[0]\n with codecs.open(path, 'r', encoding='utf-8') as f:\n lines = f.read().splitlines()\n rank=0\n resultSet=[]\n for result in lines[1:]:\n try:\n rank=rank+1\n splits=result.split('\\t')\n score = splits[0]\n trackid= splits[1]\n resultSet.append((rank,trackid,score))\n except:\n print result\n return \"QueryError\"\n return(playlistId,resultSet)\n\n####skip\nMeta1Op=[]\nerr1=[]\nMeta2Op=[]\nerr2=[]\nfor f in Meta1Files:\n res=parseMetaFiles(f)\n if res !=\"QueryError\":\n Meta1Op.append(res)\n else:\n err1.append(f)\nfor f in Meta2Files:\n res=parseMetaFiles(f)\n if res !=\"QueryError\":\n Meta2Op.append(res)\n else:\n err2.append(f)\n \n####skip\nimport codecs\ndef QEPRFParse(path):\n playlistId=path.split('/')[-1].split('.op')[0]\n with codecs.open(path, 'r', encoding='utf-8') as f:\n lines = f.read().splitlines()\n inputQueries=lines[0].split('# query: ')[1].split()\n resultSet=[]\n pairResults= lines[1].split(' #weight(')[2].split(') )')[0].split('\" ')\n rank=0\n for result in pairResults[:-1]:\n try:\n rank=rank+1\n splits=result.split('\"')\n score = splits[0].strip()\n trackid= splits[1].strip()\n resultSet.append((rank,trackid,score))\n except:\n print result\n return \"QueryError\"\n return(playlistId,inputQueries,resultSet)\n\n###skip\nQEPRFOp=[]\nerr3=[]\nfor f in QEPRFFiles:\n res=QEPRFParse(f)\n if res !=\"QueryError\":\n QEPRFOp.append(res)\n else:\n err3.append(f)\n \n###skip\nimport pickle\npidTrackMapping=pickle.load(open('./BiPartites/AllDataPidTrackListBipartite.pkl','rb'))\n\n####skip\nimport pickle\nimport os\nimport codecs\nfrom random import shuffle\n\npkl = os.listdir('./SplitsInformation/')\n\ncount=0\nDS={}\nfor fpkl in pkl:\n if fpkl in ['testing25RandPid.pkl', 'testing25Pid.pkl', 'testing1Pid.pkl', 'testing100Pid.pkl', 'testing10Pid.pkl', 'testing5Pid.pkl', 'testing100RandPid.pkl']:\n testType=fpkl.replace('.pkl','')\n if 'Rand' in fpkl:\n listLen=int(fpkl.split('testing')[1].split('Rand')[0])\n qtype='Rand'\n else :\n listLen=int(fpkl.split('testing')[1].split('Pid')[0])\n qtype='Normal'\n testingPids=pickle.load(open('./SplitsInformation/'+fpkl,'rb'))\n for pid in testingPids:\n pid=str(pid)\n referenceSet=[x.replace('spotify:track:','') for x in pidTrackMapping[pid]]\n DS[pid]=(testType,qtype,listLen,referenceSet)\n\n####skip\nimport pickle\nimport os\nimport codecs\nfrom random import shuffle\n\npkl = os.listdir('./SplitsInformation/')\ntestingTitleonlyPids=[]\nfor fpkl in pkl:\n if fpkl =='testingOnlyTitlePid.pkl':\n testType=fpkl.replace('.pkl','')\n listLen=0\n qtype='Normal'\n testingPids=pickle.load(open('./SplitsInformation/'+fpkl,'rb'))\n for pid in testingPids:\n pid=str(pid)\n referenceSet=[x.replace('spotify:track:','') for x in pidTrackMapping[pid]]\n DS[pid]=(testType,qtype,listLen,referenceSet)\ntestingTitleonlyPids=[str(x) for x in testingPids]\n\n", "# EXCEPTION in query 759505: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 864483: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 110994: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 591668: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 754459: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 479140: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 221635: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 841212: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 943553: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 512541: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 841351: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 609540: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 603765: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 593650: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 759505: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 864483: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 110994: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 591668: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 754459: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 479140: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 221635: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 841212: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 943553: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 512541: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 841351: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 609540: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 603765: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n# EXCEPTION in query 593650: IndriRunQuery.cpp(410): QueryThread::_runQuery Exception\n" ], [ "from collections import defaultdict\nfrom random import shuffle\n\nfor comb in paramSweep:\n scoresfile= comb[2]\n with open(scoresfile) as f:\n scores = f.readlines()\n \n PidTracksScores={}\n for l in scores:\n pid=l.split()[0].strip()\n trackScore=l.split()[2].strip()\n PidTracksScores.setdefault(pid,[]).append(float(trackScore))\n \n \n rerankedCandidates={}\n for pid,tracksList in PidTestTracks.items():\n scoresList=PidTracksScores[pid]\n zippedPairs=zip(tracksList,scoresList)\n shuffle(zippedPairs)\n rerankedCandidates[pid]=[x[0] for x in sorted(zippedPairs, key=lambda x: x[1], reverse=True)]\n \n ####continue here\n evalSets=[]\n for pl in QEPRFOp:\n plId=pl[0]\n exposed=pl[1]\n candidates=rerankedCandidates[plId]\n candidates=[x for x in candidates if x not in exposed]\n refVals= DS[plId]\n testtype=refVals[0]\n orderType=refVals[1]\n exposedLen=refVals[2]\n playlist=refVals[3]\n if orderType=='Normal':\n groundTruth=playlist[exposedLen:]\n else:\n groundTruth=[x for x in playlist if x not in exposed]\n evalSets.append((groundTruth, candidates[:500], testtype, exposedLen))\n\n for pl in Meta2Op:\n plId=pl[0]\n if plId in testingTitleonlyPids and plId in rerankedCandidates:\n exposed=[]\n candidates=rerankedCandidates[plId]\n refVals= DS[plId]\n testtype=refVals[0]\n orderType=refVals[1]\n exposedLen=refVals[2]\n playlist=refVals[3]\n groundTruth=playlist[exposedLen:]\n evalSets.append((groundTruth, candidates[:500], testtype, exposedLen))\n ####continue here\n\n '''\n r_precision(targets, predictions, k),\n ndcg(targets, predictions, k),\n playlist_extender_clicks(targets, predictions, k)\n '''\n\n indivSumsCounts= defaultdict(int)\n indivSumsRecall = defaultdict(int)\n indivSumsNdcg = defaultdict(int)\n indivSumsRprec = defaultdict(int)\n indivSumsClicks = defaultdict(int)\n globalNdcg=0\n globalRprec=0\n globalClicks=0\n globalRecall=0\n \n count=0\n for evalTuple in evalSets:\n targets=evalTuple[0]\n predictions=evalTuple[1]\n testType=evalTuple[2]\n tupNdcg=ndcg(targets,predictions,500)\n tuprprec=r_precision(targets,predictions,500)\n tupClicks=playlist_extender_clicks(targets,predictions,500)\n globalNdcg+=tupNdcg\n indivSumsNdcg[testType]+=tupNdcg\n globalRprec+=tuprprec\n indivSumsRprec[testType]+=tuprprec\n globalClicks+=tupClicks\n indivSumsClicks[testType]+=tupClicks\n\n indivSumsCounts[testType]+=1\n\n recallSetSize= len(set(predictions)&set(targets))\n refSetSize=len(targets)\n recall=recallSetSize*1.0/refSetSize\n globalRecall+=recall\n indivSumsRecall[testType]+=recall\n\n count+=1\n for k, v in indivSumsCounts.items():\n indivSumsRecall[k]=indivSumsRecall[k]/v\n indivSumsNdcg[k]=indivSumsNdcg[k]/v\n indivSumsRprec[k]=indivSumsRprec[k]/v\n indivSumsClicks[k]=indivSumsClicks[k]/v\n \n print scoresfile , 'Recall:' , globalRecall/count,'NDCG:', globalNdcg/count, 'RPrec:', globalRprec/count,'Clicks:', globalClicks/count\n ", "./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr09-leaf5.txt Recall: 0.496151328292 NDCG: 0.301458189779 RPrec: 0.128395610746 Clicks: 4.9145316253\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr09-leaf10.txt Recall: 0.500125251481 NDCG: 0.303755406247 RPrec: 0.129254491629 Clicks: 4.83206565252\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr09-leaf50.txt Recall: 0.504065314042 NDCG: 0.305168030632 RPrec: 0.128284880919 Clicks: 4.72037630104\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr09-leaf100.txt Recall: 0.504055750427 NDCG: 0.303409031112 RPrec: 0.126503903426 Clicks: 4.77862289832\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr09-leaf250.txt Recall: 0.499622453926 NDCG: 0.293808130357 RPrec: 0.117779772983 Clicks: 4.95756605284\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr09-leaf500.txt Recall: 0.49332748546 NDCG: 0.283637203164 RPrec: 0.110198804203 Clicks: 5.10168134508\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr05-leaf5.txt Recall: 0.496164453175 NDCG: 0.301517925208 RPrec: 0.128266662223 Clicks: 4.91713370697\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr05-leaf10.txt Recall: 0.500127634339 NDCG: 0.303780567203 RPrec: 0.129224492888 Clicks: 4.83186549239\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr05-leaf50.txt Recall: 0.504065314042 NDCG: 0.30515435263 RPrec: 0.128284880919 Clicks: 4.72017614091\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr05-leaf100.txt Recall: 0.504055750427 NDCG: 0.303422657114 RPrec: 0.126503903426 Clicks: 4.77862289832\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr05-leaf250.txt Recall: 0.499622453926 NDCG: 0.293801992519 RPrec: 0.117771432978 Clicks: 4.95756605284\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr05-leaf500.txt Recall: 0.49332748546 NDCG: 0.283652023598 RPrec: 0.110198804203 Clicks: 5.10168134508\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr01-leaf5.txt Recall: 0.496165407699 NDCG: 0.301369844886 RPrec: 0.12817000256 Clicks: 4.91993594876\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr01-leaf10.txt Recall: 0.500127634339 NDCG: 0.303673536871 RPrec: 0.12900945099 Clicks: 4.83206565252\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr01-leaf50.txt Recall: 0.504065314042 NDCG: 0.305162989172 RPrec: 0.128292579385 Clicks: 4.72037630104\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr01-leaf100.txt Recall: 0.504055750427 NDCG: 0.303382816495 RPrec: 0.126503903426 Clicks: 4.77862289832\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr01-leaf250.txt Recall: 0.499622453926 NDCG: 0.293783396179 RPrec: 0.117779772983 Clicks: 4.95756605284\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr01-leaf500.txt Recall: 0.49332748546 NDCG: 0.283655799509 RPrec: 0.110198804203 Clicks: 5.10168134508\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr001-leaf5.txt Recall: 0.496152935437 NDCG: 0.301572391142 RPrec: 0.128464317926 Clicks: 4.91533226581\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr001-leaf10.txt Recall: 0.500141931491 NDCG: 0.3037733597 RPrec: 0.129199724959 Clicks: 4.83086469175\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr001-leaf50.txt Recall: 0.504065314042 NDCG: 0.305154036099 RPrec: 0.128292579385 Clicks: 4.72037630104\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr001-leaf100.txt Recall: 0.504055750427 NDCG: 0.30341355168 RPrec: 0.126503903426 Clicks: 4.77842273819\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr001-leaf250.txt Recall: 0.499622453926 NDCG: 0.293802539574 RPrec: 0.117771432978 Clicks: 4.95756605284\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr001-leaf500.txt Recall: 0.49332748546 NDCG: 0.283656416129 RPrec: 0.110198804203 Clicks: 5.10168134508\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr0001-leaf5.txt Recall: 0.496165066787 NDCG: 0.301400850531 RPrec: 0.128153690613 Clicks: 4.91653322658\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr0001-leaf10.txt Recall: 0.500120485763 NDCG: 0.30376808345 RPrec: 0.128972121528 Clicks: 4.83106485188\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr0001-leaf50.txt Recall: 0.504065314042 NDCG: 0.305138898946 RPrec: 0.128284880919 Clicks: 4.72017614091\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr0001-leaf100.txt Recall: 0.504055750427 NDCG: 0.303350668028 RPrec: 0.126503903426 Clicks: 4.77862289832\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr0001-leaf250.txt Recall: 0.499622453926 NDCG: 0.293790149638 RPrec: 0.117771432978 Clicks: 4.95756605284\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc-1-lr0001-leaf500.txt Recall: 0.49332748546 NDCG: 0.283655797676 RPrec: 0.110198804203 Clicks: 5.10168134508\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr09-leaf5.txt Recall: 0.289335874105 NDCG: 0.157248885563 RPrec: 0.0373704395215 Clicks: 10.1407125701\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr09-leaf10.txt Recall: 0.291616788317 NDCG: 0.15700164328 RPrec: 0.0371544095338 Clicks: 10.1214971978\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr09-leaf50.txt Recall: 0.290032866751 NDCG: 0.157043067529 RPrec: 0.0375357093035 Clicks: 10.1435148118\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr09-leaf100.txt Recall: 0.293505634966 NDCG: 0.158466253 RPrec: 0.0372050045938 Clicks: 9.98899119295\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr09-leaf250.txt Recall: 0.291211243576 NDCG: 0.156863875924 RPrec: 0.0368054582285 Clicks: 10.1777421938\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr09-leaf500.txt Recall: 0.290273764549 NDCG: 0.157702522635 RPrec: 0.0380846257769 Clicks: 10.2063650921\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr05-leaf5.txt Recall: 0.288963565608 NDCG: 0.156078859737 RPrec: 0.0369598227219 Clicks: 10.2135708567\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr05-leaf10.txt Recall: 0.292462374914 NDCG: 0.157396341281 RPrec: 0.0371254143317 Clicks: 10.1989591673\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr05-leaf50.txt Recall: 0.289984230325 NDCG: 0.157202004015 RPrec: 0.0372611621779 Clicks: 10.2189751801\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr05-leaf100.txt Recall: 0.290166535871 NDCG: 0.157728661532 RPrec: 0.0374146897118 Clicks: 10.2357886309\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr05-leaf250.txt Recall: 0.290899587463 NDCG: 0.157710404275 RPrec: 0.0372930430812 Clicks: 10.15632506\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr05-leaf500.txt Recall: 0.292045907732 NDCG: 0.157762590374 RPrec: 0.0373490364783 Clicks: 9.92373899119\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr01-leaf5.txt Recall: 0.289518977162 NDCG: 0.15662697437 RPrec: 0.0376159443714 Clicks: 10.1353082466\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr01-leaf10.txt Recall: 0.290570467606 NDCG: 0.156576601465 RPrec: 0.0370202337618 Clicks: 10.2415932746\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr01-leaf50.txt Recall: 0.289711990082 NDCG: 0.156976433957 RPrec: 0.0369661264375 Clicks: 10.0258206565\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr01-leaf100.txt Recall: 0.289126374553 NDCG: 0.157181360169 RPrec: 0.0373717403651 Clicks: 10.0930744596\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr01-leaf250.txt Recall: 0.291311056084 NDCG: 0.156940280153 RPrec: 0.0370540913273 Clicks: 10.0924739792\n./Training/ModelsAndScores/BRScores750-2080Model-100Trees-NDCG20-tc1-lr01-leaf500.txt Recall: 0.290188286261 NDCG: 0.157513402744 RPrec: 0.0374607698482 Clicks: 10.0040032026\n" ], [ " Recall 0.5964542020518547,\n NDCG 0.30332032798678032,\n RPrec 0.12934009424035461,\n Clicks 5.1286", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# [NTDS'19] assignment 1: network science\n[ntds'19]: https://github.com/mdeff/ntds_2019\n\n[Eda Bayram](https://lts4.epfl.ch/bayram), [EPFL LTS4](https://lts4.epfl.ch) and\n[Nikolaos Karalias](https://people.epfl.ch/nikolaos.karalias), [EPFL LTS2](https://lts2.epfl.ch).", "_____no_output_____" ], [ "## Students\n\n* Team: `<5>`\n* `<Alice Bizeul, Gaia Carparelli, Antoine Spahr and Hugues Vinzant`", "_____no_output_____" ], [ "## Rules\n\nGrading:\n* The first deadline is for individual submissions. The second deadline is for the team submission.\n* All team members will receive the same grade based on the team solution submitted on the second deadline.\n* As a fallback, a team can ask for individual grading. In that case, solutions submitted on the first deadline are graded.\n* Collaboration between team members is encouraged. No collaboration between teams is allowed.\n\nSubmission:\n* Textual answers shall be short. Typically one to two sentences.\n* Code has to be clean.\n* You cannot import any other library than we imported.\n Note that Networkx is imported in the second section and cannot be used in the first.\n* When submitting, the notebook is executed and the results are stored. I.e., if you open the notebook again it should show numerical results and plots. We won't be able to execute your notebooks.\n* The notebook is re-executed from a blank state before submission. That is to be sure it is reproducible. You can click \"Kernel\" then \"Restart Kernel and Run All Cells\" in Jupyter.", "_____no_output_____" ], [ "## Objective\n\nThe purpose of this milestone is to explore a given dataset, represent it by network by constructing different graphs. In the first section, you will analyze the network properties. In the second section, you will explore various network models and find out the network model fitting the ones you construct from the dataset.", "_____no_output_____" ], [ "## Cora Dataset\n\nThe [Cora dataset](https://linqs.soe.ucsc.edu/node/236) consists of scientific publications classified into one of seven research fields. \n\n* **Citation graph:** the citation network can be constructed from the connections given in the `cora.cites` file.\n* **Feature graph:** each publication in the dataset is described by a 0/1-valued word vector indicating the absence/presence of the corresponding word from the dictionary and its research field, given in the `cora.content` file. The dictionary consists of 1433 unique words. A feature graph can be constructed using the Euclidean distance between the feature vector of the publications.\n\nThe [`README`](data/cora/README) provides details about the content of [`cora.cites`](data/cora/cora.cites) and [`cora.content`](data/cora/cora.content).", "_____no_output_____" ], [ "## Section 1: Network Properties", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd\nfrom matplotlib import pyplot as plt\n\n%matplotlib inline", "_____no_output_____" ] ], [ [ "### Question 1: Construct a Citation Graph and a Feature Graph", "_____no_output_____" ], [ "Read the `cora.content` file into a Pandas DataFrame by setting a header for the column names. Check the `README` file.", "_____no_output_____" ] ], [ [ "column_list = ['paper_id'] + [str(i) for i in range(1,1434)] + ['class_label']\npd_content = pd.read_csv('data/cora/cora.content', delimiter='\\t', names=column_list) \npd_content.head()", "_____no_output_____" ] ], [ [ "Print out the number of papers contained in each of the reasearch fields.\n\n**Hint:** You can use the `value_counts()` function.", "_____no_output_____" ] ], [ [ "pd_content['class_label'].value_counts()", "_____no_output_____" ] ], [ [ "Select all papers from a field of your choice and store their feature vectors into a NumPy array.\nCheck its shape.", "_____no_output_____" ] ], [ [ "my_field = 'Neural_Networks'\nfeatures = pd_content[pd_content['class_label'] == my_field].drop(columns=['paper_id','class_label']).to_numpy()\nfeatures.shape", "_____no_output_____" ] ], [ [ "Let $D$ be the Euclidean distance matrix whose $(i,j)$ entry corresponds to the Euclidean distance between feature vectors $i$ and $j$.\nUsing the feature vectors of the papers from the field which you have selected, construct $D$ as a Numpy array. ", "_____no_output_____" ] ], [ [ "distance = np.zeros([features.shape[0],features.shape[0]])\n\nfor i in range(features.shape[0]):\n distance[i] = np.sqrt(np.sum((features[i,:] - features)**2, axis=1))", "_____no_output_____" ], [ "distance.shape", "_____no_output_____" ] ], [ [ "Check the mean pairwise distance $\\mathbb{E}[D]$.", "_____no_output_____" ] ], [ [ "# Mean on the upper triangle as the matrix is symetric (we also excluded the diagonal)\nmean_distance = distance[np.triu_indices(distance.shape[1],1)].mean()\nprint('Mean euclidian distance between feature vectors of papers on Neural Networks: {}'.format(mean_distance))", "Mean euclidian distance between feature vectors of papers on Neural Networks: 5.696602496555962\n" ] ], [ [ "Plot an histogram of the euclidean distances.", "_____no_output_____" ] ], [ [ "fig,ax = plt.subplots(1,1,figsize=(8, 8))\nax.hist(distance.flatten(), density=True, bins=20, color='salmon', edgecolor='black', linewidth=1);\nax.set_title(\"Histogram of Euclidean distances between Neural-networks papers\")\nax.set_xlabel(\"Euclidian Distances\")\nax.set_ylabel(\"Frequency\")\nax.grid(True, which='major', axis='y')\nax.set_axisbelow(True)\nplt.show()", "_____no_output_____" ] ], [ [ "Now create an adjacency matrix for the papers by thresholding the Euclidean distance matrix.\nThe resulting (unweighted) adjacency matrix should have entries\n$$ A_{ij} = \\begin{cases} 1, \\; \\text{if} \\; d(i,j)< \\mathbb{E}[D], \\; i \\neq j, \\\\ 0, \\; \\text{otherwise.} \\end{cases} $$\n\nFirst, let us choose the mean distance as the threshold.", "_____no_output_____" ] ], [ [ "threshold = mean_distance\nA_feature = np.where(distance < threshold, 1, 0)\nnp.fill_diagonal(A_feature,0)", "_____no_output_____" ] ], [ [ "Now read the `cora.cites` file and construct the citation graph by converting the given citation connections into an adjacency matrix.", "_____no_output_____" ] ], [ [ "cora_cites = np.genfromtxt('data/cora/cora.cites', delimiter='\\t')", "_____no_output_____" ], [ "papers = np.unique(cora_cites)\nA_citation = np.zeros([papers.size, papers.size])\nfor i in range(cora_cites.shape[0]):\n A_citation[np.where(papers==cora_cites[i,1]),np.where(papers==cora_cites[i,0])] = 1\nA_citation.shape", "_____no_output_____" ] ], [ [ "Get the adjacency matrix of the citation graph for the field that you chose.\nYou have to appropriately reduce the adjacency matrix of the citation graph.", "_____no_output_____" ] ], [ [ "# get the paper id from the chosen field\nfield_id = pd_content[pd_content['class_label'] == my_field][\"paper_id\"].unique()\n\n# get the index of those paper in the A_citation matrix (similar to index on the vector 'papers')\nfield_citation_id = np.empty(field_id.shape[0]).astype(int)\nfor i in range(field_id.shape[0]):\n field_citation_id[i] = np.where(papers == field_id[i])[0]\n\n# get the A_citation matrix only at the index of the paper in the field \nA_citation = A_citation[field_citation_id][:,field_citation_id] \nA_citation.shape", "_____no_output_____" ] ], [ [ "Check if your adjacency matrix is symmetric. Symmetrize your final adjacency matrix if it's not already symmetric.", "_____no_output_____" ] ], [ [ "# a matrix is symetric if it's the same as its transpose\nprint('The citation adjency matrix for papers on Neural Networks is symmetric: {}'.format(np.all(A_citation == A_citation.transpose())))\n# symetrize it by taking the maximum between A and A.transposed\nA_citation = np.maximum(A_citation, A_citation.transpose())\n\n# To verify if the matrix is symetric\nprint('After modifiying the matrix, it is symmetric: {}'.format(np.count_nonzero(A_citation - A_citation.transpose())==0))", "The citation adjency matrix for papers on Neural Networks is symmetric: False\nAfter modifiying the matrix, it is symmetric: True\n" ] ], [ [ "Check the shape of your adjacency matrix again.", "_____no_output_____" ] ], [ [ "A_citation.shape", "_____no_output_____" ] ], [ [ "### Question 2: Degree Distribution and Moments", "_____no_output_____" ], [ "What is the total number of edges in each graph?", "_____no_output_____" ] ], [ [ "num_edges_feature = int(np.sum(A_feature)/2) # only half of the matrix\nnum_edges_citation = int(np.sum(A_citation)/2)\nprint(f\"Number of edges in the feature graph: {num_edges_feature}\")\nprint(f\"Number of edges in the citation graph: {num_edges_citation}\")", "Number of edges in the feature graph: 136771\nNumber of edges in the citation graph: 1175\n" ] ], [ [ "Plot the degree distribution histogram for each of the graphs.", "_____no_output_____" ] ], [ [ "degrees_citation = A_citation.sum(axis=1) # degree = nbr of connections --> sum of ones over columns (axis=1)\ndegrees_feature = A_feature.sum(axis=1) \n\ndeg_hist_normalization = np.ones(degrees_citation.shape[0]) / degrees_citation.shape[0]\n\nfig, axes = plt.subplots(1, 2, figsize=(16, 8))\naxes[0].set_title('Citation graph degree distribution')\naxes[0].hist(degrees_citation, weights=deg_hist_normalization, bins=20, color='salmon', edgecolor='black', linewidth=1);\naxes[1].set_title('Feature graph degree distribution')\naxes[1].hist(degrees_feature, weights=deg_hist_normalization, bins=20, color='salmon', edgecolor='black', linewidth=1);", "_____no_output_____" ] ], [ [ "Calculate the first and second moments of the degree distribution of each graph.", "_____no_output_____" ] ], [ [ "cit_moment_1 = np.mean(degrees_citation)\ncit_moment_2 = np.var(degrees_citation)\n\nfeat_moment_1 = np.mean(degrees_feature)\nfeat_moment_2 = np.var(degrees_feature)\n\nprint(f\"1st moment of citation graph: {cit_moment_1:.3f}\")\nprint(f\"2nd moment of citation graph: {cit_moment_2:.3f}\")\nprint(f\"1st moment of feature graph: {feat_moment_1:.3f}\")\nprint(f\"2nd moment of feature graph: {feat_moment_2:.3f}\")", "1st moment of citation graph: 2.873\n2nd moment of citation graph: 15.512\n1st moment of feature graph: 334.403\n2nd moment of feature graph: 55375.549\n" ] ], [ [ "What information do the moments provide you about the graphs?\nExplain the differences in moments between graphs by comparing their degree distributions.", "_____no_output_____" ], [ "### Answer :\n**<br>The moments provide an idea of the sparsity of the graphs and the way the data is distributed using numerical values. The first moment is associated with the average value, the second to the variance of the distribution. A large 1st moment would mean a large number of edges per node on average, whereas the 2nd moment give information about the spread of the node's degree around the average value (variance).\n<br> Citation degree distribution 1st moment lays around 2.8, and the second one is higher (around 15.5) with a large number of nodes (818). It thus means that there are many nodes with a small degree but there are also larger hubs, the nework is likely to be sparse. The feature degree distribution moments are larger, meaning a rather dense graph. There are many nodes with a degree of above 800 (15%), and since the network contains 818 nodes, it means that many nodes are almost saturated. The high variance shows that the degree distribution is more diffuse around the average value than for the citation graph.**", "_____no_output_____" ], [ "Select the 20 largest hubs for each of the graphs and remove them. Observe the sparsity pattern of the adjacency matrices of the citation and feature graphs before and after such a reduction.", "_____no_output_____" ] ], [ [ "smallest_feat_hub_idx = np.argpartition(degrees_feature, degrees_feature.shape[0]-20)[:-20]\nsmallest_feat_hub_idx.sort()\nreduced_A_feature = A_feature[smallest_feat_hub_idx][:,smallest_feat_hub_idx]\n\nsmallest_cit_hub_idx = np.argpartition(degrees_citation, degrees_citation.shape[0]-20)[:-20]\nsmallest_cit_hub_idx.sort()\nreduced_A_citation = A_citation[smallest_cit_hub_idx][:,smallest_cit_hub_idx]\n\nfig, axes = plt.subplots(2, 2, figsize=(16, 16))\naxes[0, 0].set_title('Feature graph: adjacency matrix sparsity pattern')\naxes[0, 0].spy(A_feature);\naxes[0, 1].set_title('Feature graph without top 20 hubs: adjacency matrix sparsity pattern')\naxes[0, 1].spy(reduced_A_feature);\naxes[1, 0].set_title('Citation graph: adjacency matrix sparsity pattern')\naxes[1, 0].spy(A_citation);\naxes[1, 1].set_title('Citation graph without top 20 hubs: adjacency matrix sparsity pattern')\naxes[1, 1].spy(reduced_A_citation);", "_____no_output_____" ] ], [ [ "Plot the new degree distribution histograms.", "_____no_output_____" ] ], [ [ "reduced_degrees_feat = reduced_A_feature.sum(axis=1)\nreduced_degrees_cit = reduced_A_citation.sum(axis=1)\n\ndeg_hist_normalization = np.ones(reduced_degrees_feat.shape[0])/reduced_degrees_feat.shape[0]\nfig, axes = plt.subplots(1, 2, figsize=(16, 8))\naxes[0].set_title('Citation graph degree distribution')\naxes[0].hist(reduced_degrees_cit, weights=deg_hist_normalization, bins=8, color='salmon', edgecolor='black', linewidth=1);\naxes[1].set_title('Feature graph degree distribution')\naxes[1].hist(reduced_degrees_feat, weights=deg_hist_normalization, bins=20, color='salmon', edgecolor='black', linewidth=1);", "_____no_output_____" ] ], [ [ "Compute the first and second moments for the new graphs.", "_____no_output_____" ] ], [ [ "reduced_cit_moment_1 = np.mean(reduced_degrees_cit)\nreduced_cit_moment_2 = np.var(reduced_degrees_cit)\n\nreduced_feat_moment_1 = np.mean(reduced_degrees_feat)\nreduced_feat_moment_2 = np.var(reduced_degrees_feat)\n\nprint(f\"Citation graph first moment: {reduced_cit_moment_1:.3f}\")\nprint(f\"Citation graph second moment: {reduced_cit_moment_2:.3f}\")\nprint(f\"Feature graph first moment: {reduced_feat_moment_1:.3f}\")\nprint(f\"Feature graph second moment: {reduced_feat_moment_2:.3f}\")", "Citation graph first moment: 1.972\nCitation graph second moment: 2.380\nFeature graph first moment: 302.308\nFeature graph second moment: 50780.035\n" ] ], [ [ "Print the number of edges in the reduced graphs.", "_____no_output_____" ] ], [ [ "num_edges_reduced_feature = int(np.sum(reduced_A_feature)/2)\nnum_edges_reduced_citation = int(np.sum(reduced_A_citation)/2)\nprint(f\"Number of edges in the reduced feature graph: {num_edges_reduced_feature}\")\nprint(f\"Number of edges in the reduced citation graph: {num_edges_reduced_citation}\")", "Number of edges in the reduced feature graph: 120621\nNumber of edges in the reduced citation graph: 787\n" ] ], [ [ "Is the effect of removing the hubs the same for both networks? Look at the percentage changes for each moment. Which of the moments is affected the most and in which graph? Explain why. \n\n**Hint:** Examine the degree distributions.", "_____no_output_____" ] ], [ [ "change_cit_moment_1 = (reduced_cit_moment_1-cit_moment_1)/cit_moment_1\nchange_cit_moment_2 = (reduced_cit_moment_2-cit_moment_2)/cit_moment_2\nchange_feat_moment_1 = (reduced_feat_moment_1-feat_moment_1)/feat_moment_1\nchange_feat_moment_2 = (reduced_feat_moment_2-feat_moment_2)/feat_moment_2\n\nprint(f\"% Percentage of change for citation 1st moment: {change_cit_moment_1*100:.3f}\")\nprint(f\"% Percentage of change for citation 2nd moment: {change_cit_moment_2*100:.3f}\")\nprint(f\"% Percentage of change for feature 1st moment: {change_feat_moment_1*100:.3f}\")\nprint(f\"% Percentage of change for feature 2nd moment: {change_feat_moment_2*100:.3f}\")", "% Percentage of change for citation 1st moment: -31.343\n% Percentage of change for citation 2nd moment: -84.656\n% Percentage of change for feature 1st moment: -9.598\n% Percentage of change for feature 2nd moment: -8.299\n" ] ], [ [ "### Answer : \n**After looking of the percentage of change of moments, we can notice that the removal of the 20 largest hubs affects way more the citation degree distribution than the feature degree distribution. The 2nd moment of the citation degree distribution is reduced by almost 85%, this can be due to the fact that the percentage of nodes with a high degree was lower for the citation than for the feature network and they were thus all removed as part of the 20 largest hubs, resulting in a much lower variance (less spread distribution).**\n\n**In conclusion, the new citation distribution is more condensed around its mean value, the degree landscape is hence more uniform. Regarding the feature degree distribution, a consistent number of nodes remain hotspots.**", "_____no_output_____" ], [ "### Question 3: Pruning, sparsity, paths", "_____no_output_____" ], [ "By adjusting the threshold of the euclidean distance matrix, prune the feature graph so that its number of edges is roughly close (within a hundred edges) to the number of edges in the citation graph.", "_____no_output_____" ] ], [ [ "threshold = np.max(distance)\ndiagonal = distance.shape[0]\nthreshold_flag = False\nepsilon = 0.01*threshold \ntolerance = 250\n\nwhile threshold > 0 and not threshold_flag:\n threshold -= epsilon # steps of 1% of maximum\n n_edge = int((np.count_nonzero(np.where(distance < threshold, 1, 0)) - diagonal)/2)\n \n # within a hundred edges\n if abs(num_edges_citation - n_edge) < tolerance:\n threshold_flag = True\n print(f'Found a threshold : {threshold:.3f}')\n\nA_feature_pruned = np.where(distance < threshold, 1, 0)\nnp.fill_diagonal(A_feature_pruned, 0)\nnum_edges_feature_pruned = int(np.count_nonzero(A_feature_pruned)/2)\n\nprint(f\"Number of edges in the feature graph: {num_edges_feature}\")\nprint(f\"Number of edges in the feature graph after pruning: {num_edges_feature_pruned}\")\nprint(f\"Number of edges in the citation graph: {num_edges_citation}\")", "Found a threshold : 2.957\nNumber of edges in the feature graph: 136771\nNumber of edges in the feature graph after pruning: 1386\nNumber of edges in the citation graph: 1175\n" ] ], [ [ "### Remark:\n**The distribution of distances (which is a distribution of integers) for this particular field (Neural Networks) doesn't allow a configuration where the number of edges is roughly close (whithin a hundred of edges) to the citation distribution. This is independant of the chosen epsilon . The closest match is 250 edges apart.**", "_____no_output_____" ], [ "Check your results by comparing the sparsity patterns and total number of edges between the graphs.", "_____no_output_____" ] ], [ [ "fig, axes = plt.subplots(1, 2, figsize=(12, 6))\naxes[0].set_title('Citation graph sparsity')\naxes[0].spy(A_citation);\naxes[1].set_title('Feature graph sparsity')\naxes[1].spy(A_feature_pruned);", "_____no_output_____" ] ], [ [ "Let $C_{k}(i,j)$ denote the number of paths of length $k$ from node $i$ to node $j$. \n\nWe define the path matrix $P$, with entries\n$ P_{ij} = \\displaystyle\\sum_{k=0}^{N}C_{k}(i,j). $", "_____no_output_____" ], [ "Calculate the path matrices for both the citation and the unpruned feature graphs for $N =10$. \n\n**Hint:** Use [powers of the adjacency matrix](https://en.wikipedia.org/wiki/Adjacency_matrix#Matrix_powers).", "_____no_output_____" ] ], [ [ "def path_matrix(A, N=10):\n \"\"\"Compute the path matrix for matrix A for N power \"\"\"\n power_A = [A]\n for i in range(N-1):\n power_A.append(np.matmul(power_A[-1], A))\n\n return np.stack(power_A, axis=2).sum(axis=2)", "_____no_output_____" ], [ "path_matrix_citation = path_matrix(A_citation)\npath_matrix_feature = path_matrix(A_feature)", "_____no_output_____" ] ], [ [ "Check the sparsity pattern for both of path matrices.", "_____no_output_____" ] ], [ [ "fig, axes = plt.subplots(1, 2, figsize=(16, 9))\naxes[0].set_title('Citation Path matrix sparsity')\naxes[0].spy(path_matrix_citation);\naxes[1].set_title('Feature Path matrix sparsity')\naxes[1].spy(path_matrix_feature, vmin=0, vmax=1); #scaling the color bar", "_____no_output_____" ] ], [ [ "Now calculate the path matrix of the pruned feature graph for $N=10$. Plot the corresponding sparsity pattern. Is there any difference?", "_____no_output_____" ] ], [ [ "path_matrix_pruned = path_matrix(A_feature_pruned)\n\nplt.figure(figsize=(12, 6))\nplt.title('Feature Path matrix sparsity')\nplt.spy(path_matrix_pruned);", "_____no_output_____" ] ], [ [ "### Your answer here:\n<br> **Many combinations of nodes have a path matrix value of zero now, meaning that they are not within the reach of N = 10 nodes from another node. This makes sense as many edges were removed in the pruning procedure (from 136000 to 1400). Hence, the number of possible paths from i to j was reduced, reducing at the same time the amount of paths of size N. The increase of the sparsity of the adjency matrix increases the diameter of a network.**", "_____no_output_____" ], [ "Describe how you can use the above process of counting paths to determine whether a graph is connected or not. Is the original (unpruned) feature graph connected?", "_____no_output_____" ], [ "### Answer:\n<br> **The graph is connected if all points are within the reach of others. In others words, if when increasing $N$, we are able to reach a point where the path matrix doesn't contain any null value, it means that the graph is connected. Therefore, even if the path matrix has some null values it can be connected, this depends on the chosen N-value.\n<br> For example, if 20 nodes are aligned and linked then we know that all point are reachable. Even though, the number of paths of length 10 between the first and the last node remain 0.**", "_____no_output_____" ], [ "If the graph is connected, how can you guess its diameter using the path matrix?", "_____no_output_____" ], [ "### Answer :\n<br> **The diameter coresponds to the minimum $N$ ($N$ being a non negative integer) for which the path matrix does not contain any null value.**", "_____no_output_____" ], [ "If any of your graphs is connected, calculate the diameter using that process.", "_____no_output_____" ] ], [ [ "N=0\ndiameter = None\nd_found = False \nwhile not d_found:\n N += 1\n P = path_matrix(A_feature, N)\n if np.count_nonzero(P == 0) == 0: # if there are no zero in P\n d_found = True\n diameter = N\n\nprint(f\"The diameter of the feature graph (which is connected) is: {diameter}\")", "The diameter of the feature graph (which is connected) is: 2\n" ] ], [ [ "Check if your guess was correct using [NetworkX](https://networkx.github.io/documentation/stable/reference/algorithms/generated/networkx.algorithms.distance_measures.diameter.html).\nNote: usage of NetworkX is only allowed in this part of Section 1.", "_____no_output_____" ] ], [ [ "import networkx as nx\nfeature_graph = nx.from_numpy_matrix(A_feature)\nprint(f\"Diameter of feature graph according to networkx: {nx.diameter(feature_graph)}\")", "Diameter of feature graph according to networkx: 2\n" ] ], [ [ "## Section 2: Network Models", "_____no_output_____" ], [ "In this section, you will analyze the feature and citation graphs you constructed in the previous section in terms of the network model types.\nFor this purpose, you can use the NetworkX libary imported below.", "_____no_output_____" ] ], [ [ "import networkx as nx", "_____no_output_____" ] ], [ [ "Let us create NetworkX graph objects from the adjacency matrices computed in the previous section.", "_____no_output_____" ] ], [ [ "G_citation = nx.from_numpy_matrix(A_citation)\nprint('Number of nodes: {}, Number of edges: {}'. format(G_citation.number_of_nodes(), G_citation.number_of_edges()))\nprint('Number of self-loops: {}, Number of connected components: {}'. format(G_citation.number_of_selfloops(), nx.number_connected_components(G_citation)))", "Number of nodes: 818, Number of edges: 1175\nNumber of self-loops: 0, Number of connected components: 104\n" ] ], [ [ "In the rest of this assignment, we will consider the pruned feature graph as the feature network.", "_____no_output_____" ] ], [ [ "G_feature = nx.from_numpy_matrix(A_feature_pruned)\nprint('Number of nodes: {}, Number of edges: {}'. format(G_feature.number_of_nodes(), G_feature.number_of_edges()))\nprint('Number of self-loops: {}, Number of connected components: {}'. format(G_feature.number_of_selfloops(), nx.number_connected_components(G_feature)))", "Number of nodes: 818, Number of edges: 1386\nNumber of self-loops: 0, Number of connected components: 684\n" ] ], [ [ "### Question 4: Simulation with Erdős–Rényi and Barabási–Albert models", "_____no_output_____" ], [ "Create an Erdős–Rényi and a Barabási–Albert graph using NetworkX to simulate the citation graph and the feature graph you have. When choosing parameters for the networks, take into account the number of vertices and edges of the original networks.", "_____no_output_____" ], [ "The number of nodes should exactly match the number of nodes in the original citation and feature graphs.", "_____no_output_____" ] ], [ [ "assert len(G_citation.nodes()) == len(G_feature.nodes())\nn = len(G_citation.nodes())\nprint('The number of nodes ({}) matches the original number of nodes: {}'.format(n,n==A_citation.shape[0]))", "The number of nodes (818) matches the original number of nodes: True\n" ] ], [ [ "The number of match shall fit the average of the number of edges in the citation and the feature graph.", "_____no_output_____" ] ], [ [ "m = np.round((G_citation.size() + G_feature.size()) / 2)\nprint('The number of match ({}) fits the average number of edges: {}'.format(m,m==np.round(np.mean([num_edges_citation,num_edges_feature_pruned]))))", "The number of match (1280.0) fits the average number of edges: True\n" ] ], [ [ "How do you determine the probability parameter for the Erdős–Rényi graph?", "_____no_output_____" ], [ "### Answer:\n**<br>Based on the principles ruling random networks (no preferential attachment but a random attachment), the expected number of edges is given by : $\\langle L \\rangle = p\\frac{N(N-1)}{2}$ , where $\\langle L \\rangle$ is the average number of edges, $N$, the number of nodes and $p$, the probability parameter.\n<br> Therefore we can get $p$ from the number of edges we want and the number of nodes we have : $ p = \\langle L \\rangle\\frac{2}{N(N-1)}$ \n<br> The number of expected edges is given by $m$ in our case (defined as being the average of edges between the two original networks). $N$ is the same as in the original graphs**", "_____no_output_____" ] ], [ [ "p = m*2/(n*(n-1))\nG_er = nx.erdos_renyi_graph(n, p)", "_____no_output_____" ] ], [ [ "Check the number of edges in the Erdős–Rényi graph.", "_____no_output_____" ] ], [ [ "print('My Erdos-Rényi network that simulates the citation graph has {} edges.'.format(G_er.size()))", "My Erdos-Rényi network that simulates the citation graph has 1238 edges.\n" ] ], [ [ "How do you determine the preferential attachment parameter for Barabási–Albert graphs?", "_____no_output_____" ], [ "### Answer :\n<br>**The Barabasi-Albert model uses growth and preferential attachement to build a scale-free network. The network is constructed by progressivly adding nodes to the network and adding a fixed number of edges, q, to each node added. Those edges are preferentially drawn towards already existing nodes with a high degree (preferential attachment). \n<br> By the end of the process, the network contains n nodes and hence $n * q$ edges. Knowing that the final number of edges is defined by $m$, the parameter $q = m/n$**", "_____no_output_____" ] ], [ [ "q = int(m/n)\nG_ba = nx.barabasi_albert_graph(n, q)", "_____no_output_____" ] ], [ [ "Check the number of edges in the Barabási–Albert graph.", "_____no_output_____" ] ], [ [ "print('My Barabási-Albert network that simulates the citation graph has {} edges.'.format(G_ba.size()))", "My Barabási-Albert network that simulates the citation graph has 817 edges.\n" ] ], [ [ "### Question 5: Giant Component", "_____no_output_____" ], [ "Check the size of the largest connected component in the citation and feature graphs.", "_____no_output_____" ] ], [ [ "giant_citation = max(nx.connected_component_subgraphs(G_citation), key=len)\nprint('The giant component of the citation graph has {} nodes and {} edges.'.format(giant_citation.number_of_nodes(), giant_citation.size()))", "The giant component of the citation graph has 636 nodes and 1079 edges.\n" ], [ "giant_feature = max(nx.connected_component_subgraphs(G_feature), key=len)\nprint('The giant component of the feature graph has {} nodes and {} edges.'.format(giant_feature.number_of_nodes(), giant_feature.size()))", "The giant component of the feature graph has 117 nodes and 1364 edges.\n" ] ], [ [ "Check the size of the giant components in the generated Erdős–Rényi graph.", "_____no_output_____" ] ], [ [ "giant_er = max(nx.connected_component_subgraphs(G_er), key=len)\nprint('The giant component of the Erdos-Rényi network has {} nodes and {} edges.'.format(giant_er.number_of_nodes(), giant_er.size()))", "The giant component of the Erdos-Rényi network has 771 nodes and 1234 edges.\n" ] ], [ [ "Let us match the number of nodes in the giant component of the feature graph by simulating a new Erdős–Rényi network.\nHow do you choose the probability parameter this time? \n\n**Hint:** Recall the expected giant component size from the lectures.", "_____no_output_____" ], [ "### Answer :\n**<br> We can see the average degree of a network/each node as being the probability p multiplied by the amount of nodes to which it can connect ($N-1$, because the network is not recursive, N the number of nodes,): $\\langle k \\rangle = p . (N-1)$\n<br>We can establish that, $S$, the portion of nodes in the Giant Component is given by $S = \\frac{N_{GC}}{N}$ ($N_{GC}$, the number of nodes in the giant component) and $u$, the probability that $i$ is not linked to the GC via any other node $j$. U is also the portion of nodes not in the GC : $u = 1 - S$.\n<br>Knowing that for one $j$ among the $N-1$ nodes, this probability can be seen as the probability to have no link with $j$ if $j$ is in the GC or having a link with $j$ if $j$ is not being in the GC ($p . u$), we can establish that: $u = (1 - p - p.u)^{N-1}$\n<br> Using the relationship mentionned above : $p = \\frac{<k>}{(N-1)}$, applying a log on both sides of the relationship and taking the Taylor expansion : \n<br>$S = 1-e^{-\\langle k \\rangle S}$\n<br> => $e^{-\\langle k \\rangle S} =1-S$\n<br> => $-\\langle k \\rangle S = ln(1-S)$\n<br> => $\\langle k \\rangle = -\\frac{1}{S}ln(1-S)$\n<br> This expression of the average degree is then used to define $p$ : $p = \\frac{\\langle k \\rangle}{N-1} = \\frac{-\\frac{1}{S}ln(1-S)}{N-1}$**", "_____no_output_____" ] ], [ [ "GC_node = giant_feature.number_of_nodes()\nS = GC_node/n\navg_k = -1/S*np.log(1-S)", "_____no_output_____" ], [ "p_new = avg_k/(n-1)\nG_er_new = nx.erdos_renyi_graph(n, p_new)", "_____no_output_____" ] ], [ [ "Check the size of the new Erdős–Rényi network and its giant component.", "_____no_output_____" ] ], [ [ "print('My new Erdos Renyi network that simulates the citation graph has {} edges.'.format(G_er_new.size()))\ngiant_er_new = max(nx.connected_component_subgraphs(G_er_new), key=len)\nprint('The giant component of the new Erdos-Rényi network has {} nodes and {} edges.'.format(giant_er_new.number_of_nodes(), giant_er_new.size()))", "My new Erdos Renyi network that simulates the citation graph has 437 edges.\nThe giant component of the new Erdos-Rényi network has 208 nodes and 210 edges.\n" ] ], [ [ "### Question 6: Degree Distributions", "_____no_output_____" ], [ "Recall the degree distribution of the citation and the feature graph.", "_____no_output_____" ] ], [ [ "fig, axes = plt.subplots(1, 2, figsize=(15, 6),sharex = True)\naxes[0].set_title('Citation graph')\ncitation_degrees = [deg for (node, deg) in G_citation.degree()]\naxes[0].hist(citation_degrees, bins=20, color='salmon', edgecolor='black', linewidth=1);\naxes[1].set_title('Feature graph')\nfeature_degrees = [deg for (node, deg) in G_feature.degree()]\naxes[1].hist(feature_degrees, bins=20, color='salmon', edgecolor='black', linewidth=1);", "_____no_output_____" ] ], [ [ "What does the degree distribution tell us about a network? Can you make a prediction on the network model type of the citation and the feature graph by looking at their degree distributions?", "_____no_output_____" ], [ "### Answer :\n\n<br> **The degree distribution tell us about the sparsity of a network.\nBoth show a power law degree distribution (many nodes with few edges but a couple of big components with a lot of edges). Hence they should fall in the scale-free network category which have a similar degree distribution. Therefore the Barabasi-Albert model which is a random scale free model is probably the best match.\n<br> Those distributions are indeed power laws as it could be seen by the linear behavior of the distribution using a log scale.**", "_____no_output_____" ], [ "Now, plot the degree distribution historgrams for the simulated networks.", "_____no_output_____" ] ], [ [ "fig, axes = plt.subplots(1, 3, figsize=(20, 8))\naxes[0].set_title('Erdos-Rényi network')\ner_degrees = [deg for (node, deg) in G_er.degree()]\naxes[0].hist(er_degrees, bins=10, color='salmon', edgecolor='black', linewidth=1)\naxes[1].set_title('Barabási-Albert network')\nba_degrees = [deg for (node, deg) in G_ba.degree()]\naxes[1].hist(ba_degrees, bins=10, color='salmon', edgecolor='black', linewidth=1)\naxes[2].set_title('new Erdos-Rényi network')\ner_new_degrees = [deg for (node, deg) in G_er_new.degree()]\naxes[2].hist(er_new_degrees, bins=6, color='salmon', edgecolor='black', linewidth=1)\nplt.show()", "_____no_output_____" ] ], [ [ "In terms of the degree distribution, is there a good match between the citation and feature graphs and the simulated networks?\nFor the citation graph, choose one of the simulated networks above that match its degree distribution best. Indicate your preference below.", "_____no_output_____" ], [ "### Answer :\n<br> **Regarding the feature network, none of the distributions above matche the range of degrees of the feature network. Also none of the above distributions, model the large portion of hotspots seen in the feature graph. <br>Regarding the citation network, the Barabasi-Albert network seem to be a good match. Indeed, the range of values as well as the power-law shape of the model is close to the distribution of the citation graph showed earlier. Hence, a scale free model seems to be the best match to model the citation network for the Neural Networks field.**", "_____no_output_____" ], [ "You can also simulate a network using the configuration model to match its degree disctribution exactly. Refer to [Configuration model](https://networkx.github.io/documentation/stable/reference/generated/networkx.generators.degree_seq.configuration_model.html#networkx.generators.degree_seq.configuration_model).\n\nLet us create another network to match the degree distribution of the feature graph. ", "_____no_output_____" ] ], [ [ "G_config = nx.configuration_model(feature_degrees) \nprint('Configuration model has {} nodes and {} edges.'.format(G_config.number_of_nodes(), G_config.size()))", "Configuration model has 818 nodes and 1386 edges.\n" ] ], [ [ "Does it mean that we create the same graph with the feature graph by the configuration model? If not, how do you understand that they are not the same?", "_____no_output_____" ], [ "### Answer :\n<br> **No we don't create the same graph, the number of edges, nodes and degree distribution can be the same but the links can be different. For example, in a group of three papers, various configurations are possible using only 2 links.\n<br> Also the function used to create this model considers self loops and paralell edges which is not the case for the real feature graph. Hence the network resulting from this modelisation will most probably not be identical to the original graph.**", "_____no_output_____" ], [ "### Question 7: Clustering Coefficient", "_____no_output_____" ], [ "Let us check the average clustering coefficient of the original citation and feature graphs. ", "_____no_output_____" ] ], [ [ "nx.average_clustering(G_citation)", "_____no_output_____" ], [ "nx.average_clustering(G_feature)", "_____no_output_____" ] ], [ [ "What does the clustering coefficient tell us about a network? Comment on the values you obtain for the citation and feature graph.", "_____no_output_____" ], [ "### Answer :\n**<br>Clustering coefficient is linked to the presence of subgroups (or clusters) in the network. A high clustering coefficient means that a node is very likely to be part of a subgroup. Here we can observe that the clustering coefficient of the citation graph is higher (almost double) than the one of the feature graph, this can highlight the fact that citations are more likely to form subgroups than feature.**\n", "_____no_output_____" ], [ "Now, let us check the average clustering coefficient for the simulated networks.", "_____no_output_____" ] ], [ [ "nx.average_clustering(G_er)", "_____no_output_____" ], [ "nx.average_clustering(G_ba)", "_____no_output_____" ], [ "nx.average_clustering(nx.Graph(G_config))", "_____no_output_____" ] ], [ [ "Comment on the values you obtain for the simulated networks. Is there any good match to the citation or feature graph in terms of clustering coefficient?", "_____no_output_____" ], [ "### Answer :\n<br> **No, there is not any match. The clustering coefficients are rather small compared to the ones for feature and citation graphs. Random networks have generally small clustering coefficients because they don't tend to form subgroups as the pairing is random.**", "_____no_output_____" ], [ "Check the other [network model generators](https://networkx.github.io/documentation/networkx-1.10/reference/generators.html) provided by NetworkX. Which one do you predict to have a better match to the citation graph or the feature graph in terms of degree distribution and clustering coefficient at the same time? Justify your answer.", "_____no_output_____" ], [ "### Answer :\n<br> **Based on the course notes about Watts Strogatz model which is a extension of the random network model generating small world properties and high clustering, we tested the watts_strogatz_graph function provided by NetworkX. We used the average degree ($k = m*2/n$) as an initial guess of the number of nearest neighbours to which we connect each node. We then modulated the rewiring probability to find a good match. Results did not show any satisfying match for the clustering coefficient (it was always rather low compared to the original networks). We then tuned parameter k by increasing it for a fixed p of 0.5 (corresponding to the small-world property). k was originally very low and we wanted to increase the occurence of clusters. At k =100, the clustering coefficient matched our expectations (being close to the clustering coeffients of the two distributions) but the distribtion did not match a powerlaw. In conclusion, watts-strogratz was left aside as no combination of parameters enabled to match the clustering coefficent as well as the shape of the distribution.\n<br>After scrolling through the documention of NetworkX, we came across the power_law_cluster function. According to the documentation, parameter n is the number of nodes, (n = 818 in our case). The second parameter, k, is the _number of random edges to add to each new node_ which we chose to be the average degree of the original graph as an initial guess. Parameter p, the probability of connecting two nodes which already share a common neighbour (forming a triangle) is chosen to be the average of average clustering coefficient across the original distributions. This yield a clustering coefficient that was a bit low compared with our expectations. We therefore tuned this parameter to better match the coefficients, results showed that a good comprise was reached at p = 0.27.** ", "_____no_output_____" ], [ "If you find a better fit, create a graph object below for that network model. Print the number of edges and the average clustering coefficient. Plot the histogram of the degree distribution.", "_____no_output_____" ] ], [ [ "k = m*2/n\np = (nx.average_clustering(G_citation) + nx.average_clustering(G_feature))*0.8\nG_pwc = nx.powerlaw_cluster_graph(n, int(k), p)\n \nprint('Power law cluster model has {} edges.'.format(G_pwc.size()))\nprint('Power law cluster model has a clustering coefficient of {}'.format(nx.average_clustering(G_pwc)))\nprint('Citation model has {} edges.'.format(G_citation.size()))\nprint('Citation model has a clustering coefficient of {}'.format(nx.average_clustering(G_citation)))\nprint('Feature model has {} edges.'.format(G_feature.size()))\nprint('Feature model has a clustering coefficient of {}'.format(nx.average_clustering(G_feature)))\n\nfig, axs = plt.subplots(1, 3, figsize=(15, 5), sharey=True, sharex=True)\naxs[0].set_title('PWC graph')\nws_degrees = [deg for (node, deg) in G_pwc.degree()]\naxs[0].hist(ws_degrees, bins=20, color='salmon', edgecolor='black', linewidth=1)\naxs[1].set_title('Citation graph')\ncitation_degrees = [deg for (node, deg) in G_citation.degree()]\naxs[1].hist(citation_degrees, bins=20, color='salmon', edgecolor='black', linewidth=1)\naxs[2].set_title('Feature graph')\nfeature_degree = [deg for (node, deg) in G_feature.degree()]\naxs[2].hist(feature_degree, bins=20, color='salmon', edgecolor='black', linewidth=1)\nplt.show()", "Power law cluster model has 2440 edges.\nPower law cluster model has a clustering coefficient of 0.16384898371644135\nCitation model has 1175 edges.\nCitation model has a clustering coefficient of 0.21693567980632222\nFeature model has 1386 edges.\nFeature model has a clustering coefficient of 0.1220744470334593\n" ] ], [ [ "Comment on the similarities of your match.", "_____no_output_____" ], [ "### Answer :\n<br> **At this point, the decay of the power law distribution (PWC) had an intermediate behavior between the citation and the feature graph and the clustering coefficient which fall in between the original graphs (~0.17). The degree range is roughly equilvalent in all three above distributions. \n<br> We hence found a model that satisfies both the clustering coefficients and the degree distribution. Also the final model raises distributions which have the expected intermediate behavior in comparison with the original distributions.**", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import json\nimport requests\nimport pandas as pd\nfrom geopy.distance import geodesic", "_____no_output_____" ], [ "def find_postal(lst, filename):\n \n for index,add in enumerate(lst):\n url= \"https://developers.onemap.sg/commonapi/search?returnGeom=Y&getAddrDetails=Y&pageNum=1&searchVal=\"+add\n #print(index,url)\n response = requests.get(url)\n data = json.loads(response.text) \n \n temp_df = pd.DataFrame.from_dict(data[\"results\"])\n temp_df[\"address\"] = add\n \n if index == 0:\n file = temp_df\n else:\n file = file.append(temp_df)\n file.to_csv(filename + '.csv')", "_____no_output_____" ], [ "cc = pd.read_csv('.\\Amenities\\CC.csv')\nsupermarket = pd.read_csv('.\\Amenities\\supermarket_cleaned.csv')\nhawker = pd.read_csv('.\\Amenities\\list-of-government-markets-hawker-centres.csv')\nmrt = pd.read_csv('.\\Amenities\\mrtsg.csv')\nnpc = pd.read_csv('.\\Amenities\\PoliceCentre.csv')\n", "_____no_output_____" ], [ "unique_address_cc = cc['CC'].unique()\nunique_address_supermarket = supermarket['address'].unique()\nunique_address_hawker = hawker['name_of_centre'].unique()\nunique_address_mrt = mrt['STN_NAME'].unique()\nunique_address_npc = npc['NPC'].unique()\n", "_____no_output_____" ], [ "find_postal(unique_address_supermarket, \"supermarket_geoloc\")\nfind_postal(unique_address_hawker, \"hawker_geoloc\")\nfind_postal(unique_address_mrt , \"mrt_geoloc\")\nfind_postal(unique_address_cc, \"cc_geoloc\")\nfind_postal(unique_address_supermarket, \"npc_geoloc\")", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd\ndata = pd.read_csv('input.txt', sep=\" \", header=None)\n\n\n", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Implementing a CGAN for the Iris data set to generate synthetic data\n### Import necessary modules and packages", "_____no_output_____" ] ], [ [ "import os\nwhile os.path.basename(os.getcwd()) != 'Synthetic_Data_GAN_Capstone':\n os.chdir('..')\nfrom utils.utils import *\nsafe_mkdir('experiments')\nfrom utils.data_loading import load_raw_dataset\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport torch\nimport torch.nn as nn\nimport torch.optim as optim\nfrom torch.autograd import Variable\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.preprocessing import StandardScaler\nfrom models.VGAN import VGAN_Generator, VGAN_Discriminator\nfrom models.CGAN_iris import CGAN_Generator, CGAN_Discriminator\nimport random", "_____no_output_____" ] ], [ [ "### Set random seed for reproducibility", "_____no_output_____" ] ], [ [ "manualSeed = 999\nprint(\"Random Seed: \", manualSeed)\nrandom.seed(manualSeed)\ntorch.manual_seed(manualSeed)", "Random Seed: 999\n" ] ], [ [ "### Import and briefly inspect data", "_____no_output_____" ] ], [ [ "iris = load_raw_dataset('iris')\niris.head()", "_____no_output_____" ] ], [ [ "### Preprocessing data\nSplit 50-50 so we can demonstrate the effectiveness of additional data", "_____no_output_____" ] ], [ [ "x_train, x_test, y_train, y_test = train_test_split(iris.drop(columns='species'), iris.species, test_size=0.5, stratify=iris.species, random_state=manualSeed)\nprint(\"x_train:\", x_train.shape)\nprint(\"x_test:\", x_test.shape)", "x_train: (75, 4)\nx_test: (75, 4)\n" ] ], [ [ "### Model parameters (feel free to play with these)", "_____no_output_____" ] ], [ [ "nz = 32 # Size of generator noise input\nH = 16 # Size of hidden network layer\nout_dim = x_train.shape[1] # Size of output\nbs = x_train.shape[0] # Full data set\nnc = 3 # 3 different types of label in this problem\nnum_batches = 1\nnum_epochs = 10000\nexp_name = 'experiments/iris_1x16'\nsafe_mkdir(exp_name)", "_____no_output_____" ] ], [ [ "### Adam optimizer hyperparameters\nI set these based on the original paper, but feel free to play with them as well.", "_____no_output_____" ] ], [ [ "lr = 2e-4\nbeta1 = 0.5\nbeta2 = 0.999", "_____no_output_____" ] ], [ [ "### Set the device", "_____no_output_____" ] ], [ [ "device = torch.device(\"cuda:0\" if (torch.cuda.is_available()) else \"cpu\")", "_____no_output_____" ] ], [ [ "### Scale continuous inputs for neural networks", "_____no_output_____" ] ], [ [ "scaler = StandardScaler()\nx_train = scaler.fit_transform(x_train)\nx_train_tensor = torch.tensor(x_train, dtype=torch.float)\ny_train_dummies = pd.get_dummies(y_train)\ny_train_dummies_tensor = torch.tensor(y_train_dummies.values, dtype=torch.float)", "_____no_output_____" ] ], [ [ "### Instantiate nets", "_____no_output_____" ] ], [ [ "netG = CGAN_Generator(nz=nz, H=H, out_dim=out_dim, nc=nc, bs=bs, lr=lr, beta1=beta1, beta2=beta2).to(device)\nnetD = CGAN_Discriminator(H=H, out_dim=out_dim, nc=nc, lr=lr, beta1=beta1, beta2=beta2).to(device)", "_____no_output_____" ] ], [ [ "### Print models\nI chose to avoid using sequential mode in case I wanted to create non-sequential networks, it is more flexible in my opinion, but does not print out as nicely", "_____no_output_____" ] ], [ [ "print(netG)\nprint(netD)", "CGAN_Generator(\n (fc1): Linear(in_features=35, out_features=16, bias=True)\n (output): Linear(in_features=16, out_features=4, bias=True)\n (act): LeakyReLU(negative_slope=0.2)\n (loss_fn): BCELoss()\n)\nCGAN_Discriminator(\n (fc1): Linear(in_features=7, out_features=16, bias=True)\n (output): Linear(in_features=16, out_features=1, bias=True)\n (act): LeakyReLU(negative_slope=0.2)\n (m): Sigmoid()\n (loss_fn): BCELoss()\n)\n" ] ], [ [ "### Define labels", "_____no_output_____" ] ], [ [ "real_label = 1\nfake_label = 0", "_____no_output_____" ] ], [ [ "### Training Loop\nLook through the comments to better understand the steps that are taking place", "_____no_output_____" ] ], [ [ "print(\"Starting Training Loop...\")\nfor epoch in range(num_epochs):\n for i in range(num_batches): # Only one batch per epoch since our data is horrifically small\n # Update Discriminator\n # All real batch first\n real_data = x_train_tensor.to(device) # Format batch (entire data set in this case)\n real_classes = y_train_dummies_tensor.to(device)\n label = torch.full((bs,), real_label, device=device) # All real labels\n\n output = netD(real_data, real_classes).view(-1) # Forward pass with real data through Discriminator\n netD.train_one_step_real(output, label)\n\n # All fake batch next\n noise = torch.randn(bs, nz, device=device) # Generate batch of latent vectors\n fake = netG(noise, real_classes) # Fake image batch with netG\n label.fill_(fake_label)\n output = netD(fake.detach(), real_classes).view(-1)\n netD.train_one_step_fake(output, label)\n netD.combine_and_update_opt()\n netD.update_history()\n\n # Update Generator\n label.fill_(real_label) # Reverse labels, fakes are real for generator cost\n output = netD(fake, real_classes).view(-1) # Since D has been updated, perform another forward pass of all-fakes through D\n netG.train_one_step(output, label)\n netG.update_history()\n\n # Output training stats\n if epoch % 1000 == 0 or (epoch == num_epochs-1):\n print('[%d/%d]\\tLoss_D: %.4f\\tLoss_G: %.4f\\tD(x): %.4f\\tD(G(z)): %.4f / %.4f'\n % (epoch+1, num_epochs, netD.loss.item(), netG.loss.item(), netD.D_x, netD.D_G_z1, netG.D_G_z2))\n with torch.no_grad():\n fake = netG(netG.fixed_noise, real_classes).detach().cpu()\n netG.fixed_noise_outputs.append(scaler.inverse_transform(fake))\nprint(\"Training Complete\")", "Starting Training Loop...\n[1/10000]\tLoss_D: 1.3938\tLoss_G: 0.7742\tD(x): 0.4603\tD(G(z)): 0.4609 / 0.4611\n[1001/10000]\tLoss_D: 1.3487\tLoss_G: 0.7206\tD(x): 0.5089\tD(G(z)): 0.4883 / 0.4882\n[2001/10000]\tLoss_D: 1.3557\tLoss_G: 0.7212\tD(x): 0.5025\tD(G(z)): 0.4864 / 0.4864\n[3001/10000]\tLoss_D: 1.4118\tLoss_G: 0.6679\tD(x): 0.5029\tD(G(z)): 0.5138 / 0.5135\n[4001/10000]\tLoss_D: 1.3725\tLoss_G: 0.7053\tD(x): 0.5021\tD(G(z)): 0.4943 / 0.4942\n[5001/10000]\tLoss_D: 1.3690\tLoss_G: 0.7011\tD(x): 0.5064\tD(G(z)): 0.4967 / 0.4966\n[6001/10000]\tLoss_D: 1.3813\tLoss_G: 0.6954\tD(x): 0.5028\tD(G(z)): 0.4993 / 0.4994\n[7001/10000]\tLoss_D: 1.3799\tLoss_G: 0.6996\tD(x): 0.5006\tD(G(z)): 0.4970 / 0.4969\n[8001/10000]\tLoss_D: 1.3810\tLoss_G: 0.6978\tD(x): 0.5006\tD(G(z)): 0.4976 / 0.4978\n[9001/10000]\tLoss_D: 1.3871\tLoss_G: 0.6914\tD(x): 0.5008\tD(G(z)): 0.5010 / 0.5010\n[10000/10000]\tLoss_D: 1.3838\tLoss_G: 0.6955\tD(x): 0.5003\tD(G(z)): 0.4989 / 0.4989\nTraining Complete\n" ] ], [ [ "### Output diagnostic plots tracking training progress and statistics", "_____no_output_____" ] ], [ [ "%matplotlib inline\ntraining_plots(netD=netD, netG=netG, num_epochs=num_epochs, save=exp_name)", "/home/aj/.local/lib/python3.7/site-packages/matplotlib/figure.py:445: UserWarning: Matplotlib is currently using module://ipykernel.pylab.backend_inline, which is a non-GUI backend, so cannot show the figure.\n % get_backend())\n" ], [ "plot_layer_scatters(netG, title=\"Generator\", save=exp_name)", "/home/aj/.local/lib/python3.7/site-packages/matplotlib/figure.py:445: UserWarning: Matplotlib is currently using module://ipykernel.pylab.backend_inline, which is a non-GUI backend, so cannot show the figure.\n % get_backend())\n" ], [ "plot_layer_scatters(netD, title=\"Discriminator\", save=exp_name)", "/home/aj/.local/lib/python3.7/site-packages/matplotlib/figure.py:445: UserWarning: Matplotlib is currently using module://ipykernel.pylab.backend_inline, which is a non-GUI backend, so cannot show the figure.\n % get_backend())\n" ] ], [ [ "It looks like training stabilized fairly quickly, after only a few thousand iterations. The fact that the weight norm increased over time probably means that this network would benefit from some regularization.", "_____no_output_____" ], [ "### Compare performance of training on fake data versus real data\nIn this next section, we will lightly tune two models via cross-validation. The first model will be trained on the 75 real training data examples and tested on the remaining 75 testing data examples, whereas the second set of models will be trained on different amounts of generated data (no real data involved whatsoever). We will then compare performance and plot some graphs to evaluate our CGAN.", "_____no_output_____" ] ], [ [ "y_test_dummies = pd.get_dummies(y_test)\nprint(\"Dummy columns match?\", all(y_train_dummies.columns == y_test_dummies.columns))\nx_test = scaler.transform(x_test)\nlabels_list = [x for x in y_train_dummies.columns]\nparam_grid = {'tol': [1e-9, 1e-8, 1e-7, 1e-6, 1e-5],\n 'C': [0.5, 0.75, 1, 1.25],\n 'l1_ratio': [0, 0.25, 0.5, 0.75, 1]}", "Dummy columns match? True\n" ] ], [ [ "### Train on real data", "_____no_output_____" ] ], [ [ "model_real, score_real = train_test_logistic_reg(x_train, y_train, x_test, y_test, param_grid=param_grid, cv=5, random_state=manualSeed, labels=labels_list)", "Accuracy: 0.92\nBest Parameters: {'C': 1.25, 'l1_ratio': 0.5, 'tol': 1e-09}\n precision recall f1-score support\n\n Iris-setosa 1.00 1.00 1.00 25\nIris-versicolor 0.85 0.92 0.88 25\n Iris-virginica 0.91 0.84 0.87 25\n\n accuracy 0.92 75\n macro avg 0.92 0.92 0.92 75\n weighted avg 0.92 0.92 0.92 75\n\n[[25 0 0]\n [ 0 23 2]\n [ 0 4 21]]\n" ] ], [ [ "### Train on various levels of fake data", "_____no_output_____" ] ], [ [ "test_range = [75, 150, 300, 600, 1200]\nfake_bs = bs\nfake_models = []\nfake_scores = []\nfor size in test_range:\n num_batches = size // fake_bs + 1\n genned_data = np.empty((0, out_dim))\n genned_labels = np.empty(0)\n rem = size\n while rem > 0:\n curr_size = min(fake_bs, rem)\n noise = torch.randn(curr_size, nz, device=device)\n fake_labels, output_labels = gen_labels(size=curr_size, num_classes=nc, labels_list=labels_list)\n fake_labels = fake_labels.to(device)\n rem -= curr_size\n fake_data = netG(noise, fake_labels).cpu().detach().numpy()\n genned_data = np.concatenate((genned_data, fake_data))\n genned_labels = np.concatenate((genned_labels, output_labels))\n print(\"For size of:\", size)\n model_fake_tmp, score_fake_tmp = train_test_logistic_reg(genned_data, genned_labels, x_test, y_test,\n param_grid=param_grid, cv=5, random_state=manualSeed, labels=labels_list)\n fake_models.append(model_fake_tmp)\n fake_scores.append(score_fake_tmp)", "For size of: 75\nAccuracy: 0.92\nBest Parameters: {'C': 0.5, 'l1_ratio': 0.25, 'tol': 1e-09}\n precision recall f1-score support\n\n Iris-setosa 1.00 1.00 1.00 25\nIris-versicolor 0.88 0.88 0.88 25\n Iris-virginica 0.88 0.88 0.88 25\n\n accuracy 0.92 75\n macro avg 0.92 0.92 0.92 75\n weighted avg 0.92 0.92 0.92 75\n\n[[25 0 0]\n [ 0 22 3]\n [ 0 3 22]]\nFor size of: 150\nAccuracy: 0.92\nBest Parameters: {'C': 0.5, 'l1_ratio': 1, 'tol': 1e-09}\n precision recall f1-score support\n\n Iris-setosa 1.00 1.00 1.00 25\nIris-versicolor 0.88 0.88 0.88 25\n Iris-virginica 0.88 0.88 0.88 25\n\n accuracy 0.92 75\n macro avg 0.92 0.92 0.92 75\n weighted avg 0.92 0.92 0.92 75\n\n[[25 0 0]\n [ 0 22 3]\n [ 0 3 22]]\nFor size of: 300\nAccuracy: 0.92\nBest Parameters: {'C': 0.5, 'l1_ratio': 0, 'tol': 1e-09}\n precision recall f1-score support\n\n Iris-setosa 1.00 1.00 1.00 25\nIris-versicolor 0.88 0.88 0.88 25\n Iris-virginica 0.88 0.88 0.88 25\n\n accuracy 0.92 75\n macro avg 0.92 0.92 0.92 75\n weighted avg 0.92 0.92 0.92 75\n\n[[25 0 0]\n [ 0 22 3]\n [ 0 3 22]]\nFor size of: 600\nAccuracy: 0.92\nBest Parameters: {'C': 0.5, 'l1_ratio': 0.75, 'tol': 1e-09}\n precision recall f1-score support\n\n Iris-setosa 1.00 1.00 1.00 25\nIris-versicolor 0.88 0.88 0.88 25\n Iris-virginica 0.88 0.88 0.88 25\n\n accuracy 0.92 75\n macro avg 0.92 0.92 0.92 75\n weighted avg 0.92 0.92 0.92 75\n\n[[25 0 0]\n [ 0 22 3]\n [ 0 3 22]]\nFor size of: 1200\nAccuracy: 0.9466666666666667\nBest Parameters: {'C': 0.5, 'l1_ratio': 0, 'tol': 1e-09}\n precision recall f1-score support\n\n Iris-setosa 1.00 1.00 1.00 25\nIris-versicolor 0.89 0.96 0.92 25\n Iris-virginica 0.96 0.88 0.92 25\n\n accuracy 0.95 75\n macro avg 0.95 0.95 0.95 75\n weighted avg 0.95 0.95 0.95 75\n\n[[25 0 0]\n [ 0 24 1]\n [ 0 3 22]]\n" ] ], [ [ "Well, it looks like this experiment was a success. The models trained on fake data were actually able to outperform models trained on real data, which supports the belief that the CGAN is able to understand the distribution of the data it was trained on and generate meaningful examples that can be used to add additional information to the model.", "_____no_output_____" ], [ "Let's visualize some of the distributions of outputs to get a better idea of what took place", "_____no_output_____" ] ], [ [ "iris_plot_scatters(genned_data, genned_labels, \"Fake Data\", scaler, alpha=0.5, save=exp_name) # Fake data", "_____no_output_____" ], [ "iris_plot_scatters(iris.drop(columns='species'), np.array(iris.species), \"Full Real Data Set\", alpha=0.5, save=exp_name) # All real data", "_____no_output_____" ], [ "iris_plot_densities(genned_data, genned_labels, \"Fake Data\", scaler, save=exp_name) # Fake data", "/home/aj/.local/lib/python3.7/site-packages/matplotlib/figure.py:445: UserWarning: Matplotlib is currently using module://ipykernel.pylab.backend_inline, which is a non-GUI backend, so cannot show the figure.\n % get_backend())\n" ], [ "iris_plot_densities(iris.drop(columns='species'), np.array(iris.species), \"Full Real Data Set\", save=exp_name) # All real data", "/home/aj/.local/lib/python3.7/site-packages/matplotlib/figure.py:445: UserWarning: Matplotlib is currently using module://ipykernel.pylab.backend_inline, which is a non-GUI backend, so cannot show the figure.\n % get_backend())\n" ], [ "plot_scatter_matrix(genned_data, \"Fake Data\", iris.drop(columns='species'), scaler=scaler, save=exp_name)", "_____no_output_____" ], [ "plot_scatter_matrix(iris.drop(columns='species'), \"Real Data\", iris.drop(columns='species'), scaler=None, save=exp_name)", "_____no_output_____" ] ], [ [ "Finally, I present a summary of the test results ran above", "_____no_output_____" ] ], [ [ "fake_data_training_plots(real_range=bs, score_real=score_real, test_range=test_range, fake_scores=fake_scores, save=exp_name)", "/home/aj/.local/lib/python3.7/site-packages/matplotlib/figure.py:445: UserWarning: Matplotlib is currently using module://ipykernel.pylab.backend_inline, which is a non-GUI backend, so cannot show the figure.\n % get_backend())\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import qiskit\nimport numpy as np, matplotlib.pyplot as plt\nimport sys\nsys.path.insert(1, '../')\nimport qtm.base, qtm.constant, qtm.nqubit, qtm.onequbit, qtm.fubini_study", "_____no_output_____" ], [ "num_qubits = 3\nnum_layers = 2\npsi = 2*np.random.rand(2**num_qubits)-1\npsi = psi / np.linalg.norm(psi)\nqc_origin = qiskit.QuantumCircuit(num_qubits, num_qubits)\nqc_origin.initialize(psi, range(0, num_qubits))\nthetas_origin = np.ones((num_layers*num_qubits*3))", "_____no_output_____" ], [ "qc = qc_origin.copy()\nthetas = thetas_origin.copy()\nthetas, loss_values_sgd = qtm.base.fit(\n qc, num_steps = 100, thetas = thetas, \n create_circuit_func = qtm.nqubit.u_cluster_nlayer_nqubit, \n grad_func = qtm.base.grad_loss,\n loss_func = qtm.base.loss_basis,\n optimizer = qtm.base.sgd,\n verbose = 1,\n num_layers = num_layers\n)", "Step: 100%|██████████| 100/100 [14:02<00:00, 8.43s/it]\n" ], [ "qc = qc_origin.copy()\nthetas = thetas_origin.copy()\nthetas, loss_values_adam = qtm.base.fit(\n qc, num_steps = 100, thetas = thetas, \n create_circuit_func = qtm.nqubit.u_cluster_nlayer_nqubit, \n grad_func = qtm.base.grad_loss,\n loss_func = qtm.base.loss_basis,\n optimizer = qtm.base.adam,\n verbose = 1,\n num_layers = num_layers\n)", "Step: 100%|██████████| 100/100 [11:04<00:00, 6.65s/it]\n" ], [ "qc = qc_origin.copy()\nthetas = thetas_origin.copy()\nthetas, loss_values_qfsm = qtm.base.fit(\n qc, num_steps = 100, thetas = thetas, \n create_circuit_func = qtm.nqubit.u_cluster_nlayer_nqubit, \n grad_func = qtm.base.grad_loss,\n loss_func = qtm.base.loss_basis,\n optimizer = qtm.base.qng_fubini_study,\n verbose = 1,\n num_layers = num_layers\n)", "Step: 100%|██████████| 100/100 [37:54<00:00, 22.75s/it]\n" ], [ "qc = qc_origin.copy()\nthetas = thetas_origin.copy()\nthetas, loss_values_qfim = qtm.base.fit(\n qc, num_steps = 100, thetas = thetas, \n create_circuit_func = qtm.nqubit.u_cluster_nlayer_nqubit, \n grad_func = qtm.base.grad_loss,\n loss_func = qtm.base.loss_basis,\n optimizer = qtm.base.qng_qfim,\n verbose = 1,\n num_layers = num_layers\n)", "Step: 100%|██████████| 100/100 [29:50<00:00, 17.90s/it]\n" ], [ "qc = qc_origin.copy()\nthetas = thetas_origin.copy()\nthetas, loss_values_adam_qfim = qtm.base.fit(\n qc, num_steps = 100, thetas = thetas, \n create_circuit_func = qtm.nqubit.u_cluster_nlayer_nqubit, \n grad_func = qtm.base.grad_loss,\n loss_func = qtm.base.loss_basis,\n optimizer = qtm.base.qng_adam,\n verbose = 1,\n num_layers = num_layers\n)", "Step: 100%|██████████| 100/100 [29:22<00:00, 17.62s/it]\n" ], [ "plt.plot(loss_values_sgd, label=\"SGD\")\nplt.plot(loss_values_adam, label=\"Adam\")\nplt.plot(loss_values_qfsm, label=\"QNG-QFSM\")\nplt.plot(loss_values_qfim, label=\"QNG-QFIM\")\nplt.plot(loss_values_adam_qfim, label=\"Adam QNG-QFIM\")\nplt.xlabel(\"Step\")\nplt.ylabel(\"Cost value\")\nplt.ylim(-0.05, 1.05)\nplt.legend(prop={'size': 8}, loc=1)\nplt.savefig(str(num_qubits) + '.svg', format='svg')\nplt.show()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT-0", "MIT" ]

[ "MIT-0", "MIT" ]

[ "MIT-0", "MIT" ]

[ [ [ "# Predictive Maintenance using Machine Learning on Sagemaker\n*Part 3 - Timeseries data preparation*", "_____no_output_____" ], [ "## Initialization\n---\nDirectory structure to run this notebook:\n```\nnasa-turbofan-rul-lstm\n|\n+--- data\n| |\n| +--- interim: intermediate data we can manipulate and process\n| |\n| \\--- raw: *immutable* data downloaded from the source website\n|\n+--- notebooks: all the notebooks are positionned here\n|\n+--- src: utility python modules are stored here\n```\n### Imports", "_____no_output_____" ] ], [ [ "%load_ext autoreload\n\nimport matplotlib.pyplot as plt\nimport sagemaker\nimport boto3\nimport os\nimport errno\nimport pandas as pd\nimport numpy as np\nimport seaborn as sns\nimport json\nimport sys\nimport s3fs\nimport mxnet as mx\nimport joblib", "_____no_output_____" ], [ "%matplotlib inline\n%autoreload 2\n\nsns.set_style('darkgrid')\nsys.path.append('../src')\n\nfigures = []\n\nINTERIM_DATA = '../data/interim'\nPROCESSED_DATA = '../data/processed'", "_____no_output_____" ] ], [ [ "### Loading data from the previous notebook", "_____no_output_____" ] ], [ [ "# Load data from the notebook local storage:\n%store -r reduced_train_data\n%store -r reduced_test_data\n\n# If the data are not present in the notebook local storage, we need to load them from disk:\nsuccess_msg = 'Loaded \"reduced_train_data\"'\nif 'reduced_train_data' not in locals():\n print('Nothing in notebook store, trying to load from disk.')\n try:\n local_path = '../data/interim'\n reduced_train_data = pd.read_csv(os.path.join(local_path, 'reduced_train_data.csv'))\n reduced_train_data = reduced_train_data.set_index(['unit_number', 'time'])\n \n print(success_msg)\n \n except Exception as e:\n if (e.errno == errno.ENOENT):\n print('Files not found to load train data from: you need to execute the previous notebook.')\n \nelse:\n print('Train data found in notebook environment.')\n print(success_msg)\n \nsuccess_msg = 'Loaded \"reduced_test_data\"'\nif 'reduced_test_data' not in locals():\n print('Nothing in notebook store, trying to load from disk.')\n try:\n local_path = '../data/interim'\n reduced_test_data = pd.read_csv(os.path.join(local_path, 'reduced_test_data.csv'))\n reduced_test_data = reduced_test_data.set_index(['unit_number', 'time'])\n \n print(success_msg)\n \n except Exception as e:\n if (e.errno == errno.ENOENT):\n print('Files not found to load test data from: you need to execute the previous notebook.')\n \nelse:\n print('Test data found in notebook environment.')\n print(success_msg)", "no stored variable reduced_train_data\nno stored variable reduced_test_data\nNothing in notebook store, trying to load from disk.\nLoaded \"reduced_train_data\"\nNothing in notebook store, trying to load from disk.\nLoaded \"reduced_test_data\"\n" ], [ "print(reduced_train_data.shape)\nreduced_train_data.head()", "(20631, 19)\n" ], [ "print(reduced_test_data.shape)\nreduced_test_data.head()", "(13096, 26)\n" ] ], [ [ "### Study parameters", "_____no_output_____" ] ], [ [ "sequence_length = 20", "_____no_output_____" ] ], [ [ "## Normalization\n---\n### Normalizing the training data\nFirst, we build some scalers based on the training data:", "_____no_output_____" ] ], [ [ "from sklearn import preprocessing\n\n# Isolate the columns to normalize:\nnormalized_cols = reduced_train_data.columns.difference(['true_rul', 'piecewise_rul'])\n\n# Build MinMax scalers for the features and the labels:\nfeatures_scaler = preprocessing.MinMaxScaler()\nlabels_scaler = preprocessing.MinMaxScaler()\n\n# Normalize the operational settings and sensor measurements data (our features):\nreduced_train_data['sensor_measurement_17'] = reduced_train_data['sensor_measurement_17'].astype(np.float64)\nnormalized_data = pd.DataFrame(\n features_scaler.fit_transform(reduced_train_data[normalized_cols]),\n columns=normalized_cols, \n index=reduced_train_data.index\n)\n\n# Normalizing the labels data:\nreduced_train_data['piecewise_rul'] = reduced_train_data['piecewise_rul'].astype(np.float64)\nnormalized_training_labels = pd.DataFrame(\n labels_scaler.fit_transform(reduced_train_data[['piecewise_rul']]),\n columns=['piecewise_rul'], \n index=reduced_train_data.index\n)\n\n# Join the normalized features with the RUL (label) data:\njoined_data = normalized_training_labels.join(normalized_data)\nnormalized_train_data = joined_data.reindex(columns=reduced_train_data.columns)\nnormalized_train_data['true_rul'] = reduced_train_data['true_rul']\nprint(normalized_train_data.shape)\nnormalized_train_data.head()", "(20631, 19)\n" ] ], [ [ "### Normalizing the testing data\nNext, we apply these normalizers to the testing data:", "_____no_output_____" ] ], [ [ "normalized_test_data = pd.DataFrame(\n features_scaler.transform(reduced_test_data[normalized_cols]),\n columns=normalized_cols,\n index=reduced_test_data.index\n)\n\nreduced_test_data['piecewise_rul'] = reduced_test_data['piecewise_rul'].astype(np.float64)\n\nnormalized_test_labels = pd.DataFrame(\n labels_scaler.transform(reduced_test_data[['piecewise_rul']]),\n columns=['piecewise_rul'], \n index=reduced_test_data.index\n)\n\n# Join the normalize data with the RUL data:\njoined_data = normalized_test_labels.join(normalized_test_data)\nnormalized_test_data = joined_data.reindex(columns=reduced_test_data.columns)\nnormalized_test_data['true_rul'] = reduced_test_data['true_rul']\nprint(normalized_test_data.shape)\nnormalized_test_data.head()", "(13096, 26)\n" ] ], [ [ "## Sequences generation\n---", "_____no_output_____" ] ], [ [ "from lstm_utils import generate_sequences, generate_labels, generate_training_sequences, generate_testing_sequences\n\n# Building features, target and engine unit lists:\nfeatures = normalized_train_data.columns.difference(['true_rul', 'piecewise_rul'])\ntarget = ['piecewise_rul']\nunit_list = list(set(normalized_train_data.index.get_level_values(level=0).tolist()))\n\n# Generating features and labels for the training sequences:\ntrain_sequences = generate_training_sequences(normalized_train_data, sequence_length, features, unit_list)\ntrain_labels = generate_training_sequences(normalized_train_data, sequence_length, target, unit_list)\ntest_sequences, test_labels, unit_span = generate_testing_sequences(normalized_test_data, sequence_length, features, target, unit_list)\n\n# Checking sequences shapes:\nprint('train_sequences:', train_sequences.shape)\nprint('train_labels:', train_labels.shape)\nprint('test_sequences:', test_sequences.shape)\nprint('test_labels:', test_labels.shape)", "Unit 1 test sequence ignored, not enough data points.\nUnit 22 test sequence ignored, not enough data points.\nUnit 39 test sequence ignored, not enough data points.\nUnit 85 test sequence ignored, not enough data points.\ntrain_sequences: (18631, 20, 17)\ntrain_labels: (18631, 1)\ntest_sequences: (11035, 20, 17)\ntest_labels: (11035, 1)\n" ] ], [ [ "### Visualizing the sequences\nLet's visualize the sequences we built for an engine (e.g. unit 3 in the example below) to understand what will be fed to the LSTM model:", "_____no_output_____" ] ], [ [ "from lstm_utils import plot_timestep, plot_text\n\n# We use the normalized sequences for the plot but the original data for the label (RUL) for understanding purpose:\ncurrent_unit = 1\ntmp_sequences = generate_training_sequences(normalized_train_data, sequence_length, features, [current_unit])\ntmp_labels = generate_training_sequences(reduced_train_data, sequence_length, target, [current_unit])\n\n# Initialize the graphics:\nprint('Sequences generated for unit {}:\\n'.format(current_unit))\nsns.set_style('white')\nfig = plt.figure(figsize=(35,6))\n\n# Initialize the loop:\nnb_signals = min(12, len(tmp_sequences[0][0]))\nplots_per_row = nb_signals + 3\nnb_rows = 7\nnb_cols = plots_per_row\ncurrent_row = 0\nprevious_index = -1\ntimesteps = [\n # We will plot the first 3 sequences (first timesteps fed to the LSTM model):\n 0, 1, 2, \n\n # And the last 3 ones:\n len(tmp_sequences) - 3, len(tmp_sequences) - 2, len(tmp_sequences) - 1\n]\n\n# Loops through all the timesteps we want to plot:\nfor i in timesteps:\n # We draw a vertical ellispsis for the hole in the timesteps:\n if (i - previous_index > 1):\n plot_text(fig, nb_rows, nb_cols, nb_signals, current_row, '. . .', 1, no_axis=True, main_title='', plots_per_row=plots_per_row, options={'fontsize': 32, 'rotation': 270})\n current_row += 1\n\n # Timestep column:\n previous_index = i\n plot_text(fig, nb_rows, nb_cols, nb_signals, current_row, 'T{}'.format(i), 1, no_axis=True, main_title='Timestep', plots_per_row=plots_per_row, options={'fontsize': 16})\n\n # For a given timestep, we want to loop through all the signals to plot:\n plot_timestep(nb_rows, nb_cols, nb_signals, current_row, 2, tmp_sequences[i].T, features.tolist(), plots_per_row=plots_per_row)\n\n # Then we draw an ellipsis:\n plot_text(fig, nb_rows, nb_cols, nb_signals, current_row, '. . .', nb_signals + 2, no_axis=True, main_title='', plots_per_row=plots_per_row, options={'fontsize': 32})\n\n # Finally, we show the remaining useful life at the end of the row for this timestep:\n plot_text(fig, nb_rows, nb_cols, nb_signals, current_row, int(tmp_labels[i][0]), nb_signals + 3, no_axis=False, main_title='RUL', plots_per_row=plots_per_row, options={'fontsize': 16, 'color': '#CC0000'})\n\n current_row += 1", "Sequences generated for unit 1:\n\n" ] ], [ [ "## Cleanup\n---", "_____no_output_____" ], [ "### Storing data for the next notebook", "_____no_output_____" ] ], [ [ "%store labels_scaler\n%store test_sequences\n%store test_labels\n%store unit_span\n\n#columns = normalized_train_data.columns.tolist()\n#%store columns\n#%store train_sequences\n#%store train_labels\n#%store normalized_train_data\n#%store normalized_test_data\n#%store sequence_length", "Stored 'labels_scaler' (MinMaxScaler)\nStored 'test_sequences' (ndarray)\nStored 'test_labels' (ndarray)\nStored 'unit_span' (list)\n" ] ], [ [ "### Persisting these data to disk\nThis is useful in case you want to be able to execute each notebook independantly (from one session to another) and don't want to reexecute every notebooks whenever you want to focus on a particular step. Let's start by persisting train and test sequences and the associated labels:", "_____no_output_____" ] ], [ [ "import h5py as h5\n\ntrain_data = os.path.join(PROCESSED_DATA, 'train.h5')\nwith h5.File(train_data, 'w') as ftrain:\n ftrain.create_dataset('train_sequences', data=train_sequences)\n ftrain.create_dataset('train_labels', data=train_labels)\n ftrain.close()\n\ntest_data = os.path.join(PROCESSED_DATA, 'test.h5')\nwith h5.File(test_data, 'w') as ftest:\n ftest.create_dataset('test_sequences', data=test_sequences)\n ftest.create_dataset('test_labels', data=test_labels)\n ftest.close()", "_____no_output_____" ] ], [ [ "Pushing these files to S3:", "_____no_output_____" ] ], [ [ "sagemaker_session = sagemaker.Session()\nbucket = sagemaker_session.default_bucket()\nprefix = 'nasa-rul-lstm/data'\ntrain_data_location = 's3://{}/{}'.format(bucket, prefix)\n\ns3_resource = boto3.Session().resource('s3')\ns3_resource.Bucket(bucket).Object('{}/train/train.h5'.format(prefix)).upload_file(train_data)\ns3_resource.Bucket(bucket).Object('{}/test/test.h5'.format(prefix)).upload_file(test_data)\n\n# Build the data channel and write it to disk:\ndata_channels = {'train': 's3://{}/{}/train/train.h5'.format(bucket, prefix)}\nwith open(os.path.join(PROCESSED_DATA, 'data_channels.txt'), 'w') as f:\n f.write(str(data_channels))\n\n%store data_channels", "Stored 'data_channels' (dict)\n" ] ], [ [ "Storing the other elements on disk:", "_____no_output_____" ] ], [ [ "with open(os.path.join(PROCESSED_DATA, 'unit_span.lst'), 'w') as f:\n f.write(str(unit_span))\n\n_ = joblib.dump(labels_scaler, os.path.join(PROCESSED_DATA, 'labels_scaler.joblib'))", "_____no_output_____" ] ], [ [ "### Memory cleanup", "_____no_output_____" ] ], [ [ "fig.clear()\nplt.close(fig)\n\nimport gc\n_ = gc.collect()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# О вероятности попасть под удар фигуры, поставленной случайным образом на шахматную доску\n", "_____no_output_____" ], [ "На шахматную доску случайным образом поставлены две фигуры. С какой вероятностью первая фигура бьёт вторую? В данном ноутбуке представлен расчёт этой вероятности для каждой шахматной фигуры как функции от размера доски. Рассмотрены только квадратные доски. Фигуры полагаются поставленными одновременно (обязательно стоят на разных клетках), выбор каждого из полей равновероятен.", "_____no_output_____" ], [ "Степень (валентность) вершины $v$ графа $G$ - количество рёбер графа $G$, инцидентных вершине $v$.", "_____no_output_____" ], [ "Граф ходов шахматной фигуры (далее Граф) - граф, изображающий все возможные ходы фигуры на шахматной доске - каждая вершина соответствует клетке на доске, а рёбра соответствуют возможным ходам.\nТогда валентность вершины Графа - количество полей, которые бьёт фигура, будучи поставленной на соответствующую этой вершине клетку. В целях упрощения речи далее в тексте используется формулировка \"клетка имеет валентность\", однако понятно, что валентность имеет не клетка, а соответствующая ей вершина в Графе.", "_____no_output_____" ], [ "Если событие $X$ может произойти только при выполнении одного из событий $H_1, H_2,..., H_n$, которые образуют полную группу несовместных событий, то вероятность $P(X)$ вычисляется по формуле: $$P(X) = P(H_1) \\cdot P(X|H_1) + P(H_2) \\cdot P(X|H_2) + ... + P(H_n) \\cdot P(X|H_n),$$ которая называется формулой полной вероятности.", "_____no_output_____" ], [ "Пусть событие $X_{piece}$ = «Первая *фигура (piece)* бьёт вторую на доске размера $n\\times n$», $a$ - некоторое значение валентности, $b$ - количество клеток, имеющих валентность $a$, каждая из гипотез $H_i$ = «Первая фигура стоит на клетке с валентностью $a_i$». Тогда $P(H_i) = \\frac{b_i}{n^{2}}$ в силу классического определения вероятности - это отношение количества исходов, благоприятствующих событию $H_i$, к количеству всех равновозможных исходов. $P(X_{piece}|H_i) = \\frac{a_i}{n^{2}-1}$ по той же причине - событию $X_{piece}$ при условии $H_i$ благоприятствует $a_i$ исходов (вторая фигура стоит под ударом клетки с валентностью $a_i$),а количество всех равновозможных исходов уменьшилось на единицу, так как одна из клеток уже занята первой фигурой.", "_____no_output_____" ], [ "### Импорты", "_____no_output_____" ] ], [ [ "%matplotlib inline\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom string import ascii_uppercase as alphabet", "_____no_output_____" ] ], [ [ "### Функция создания доски", "_____no_output_____" ], [ "Создание доски как пустого массива заданной формы. Также нам понадобится массив такой же формы для хранения цвета надписи на клетках.\nЗаполнение доски: 0 - чёрные клетки, 1 - белые.", "_____no_output_____" ] ], [ [ "def get_board(board_size):\n\n x, y = np.meshgrid(range(board_size), range(board_size))\n\n board = np.empty(shape=(board_size, board_size), dtype='uint8')\n text_colors = np.empty_like(board, dtype='<U5')\n\n # force left bottom corner cell to be black\n if board_size % 2 == 0:\n extra_term = 1\n else:\n extra_term = 0\n\n for i, j in zip(x.flatten(), y.flatten()):\n board[i, j] = (i + j + extra_term) % 2\n # text color should be the opposite to a cell color\n text_colors[i, j] = 'black' if board[i, j] else 'white'\n\n return board, text_colors", "_____no_output_____" ], [ "def get_valencies(piece, board):\n # Get valencies for the given piece on the given board\n valencies = np.empty_like(board)\n\n if piece == 'Pawn':\n valencies = pawn(valencies)\n\n elif piece == 'Knight':\n valencies = knight(valencies)\n\n elif piece == 'Rook':\n valencies = rook(valencies)\n\n elif piece == 'King':\n valencies = king(valencies)\n\n else:\n valencies = bishop_or_queen(piece, valencies)\n\n return valencies", "_____no_output_____" ] ], [ [ "### Функция рисования доски", "_____no_output_____" ], [ "Функция создаёт изображение шахматной доски, каждая клетка которой подписана соответствующим значением валентности.", "_____no_output_____" ] ], [ [ "def plot_board(board, text_colors, piece):\n\n board_size = np.shape(board)[0]\n\n x, y = np.meshgrid(range(board_size), range(board_size))\n\n # let figure size be dependent on the board size\n plt.figure(figsize=(3*board_size/4, 3*board_size/4))\n ax = plt.subplot(111)\n\n ax.imshow(board, cmap='gray', interpolation='none')\n\n # Display valency (degree) values\n val_board = get_valencies(piece, board)\n\n for i, j, valency, text_col in zip(x.flatten(), y.flatten(),\n val_board.flatten(),\n text_colors.flatten()):\n ax.text(i, j, str(valency), color=text_col,\n va='center', ha='center', fontsize=20)\n\n ax.set_xticks(np.arange(board_size+1)) # one tick per cell\n ax.set_xticklabels(alphabet[:board_size]) # set letters as ticklabels\n # one tick per cell\n ax.set_yticks(np.arange(board_size+1))\n # set numbers as ticklabels (upside down)\n ax.set_yticklabels(np.arange(board_size, 0, -1))\n\n ax.axis('tight') # get rid of the white spaces on the edges\n ax.set_title(piece, fontsize=30)\n plt.show()", "_____no_output_____" ] ], [ [ "## Пешка", "_____no_output_____" ], [ "### Функция, возвращающая массив валентностей пешечного Графа", "_____no_output_____" ], [ "Иллюстрирует изложенные ниже соображения.", "_____no_output_____" ] ], [ [ "def pawn(valencies):\n\n valencies[0, :] = 0 # empty horizontal line\n valencies[1:, 0] = valencies[1:, -1] = 1 # vertical edges\n valencies[1:, 1:-1] = 2\n\n return valencies", "_____no_output_____" ] ], [ [ "Рассмотрим несколько частных случаев в поисках закономерности.", "_____no_output_____" ] ], [ [ "def special_cases(piece, board_sizes):\n ''' Plot boards of every board_size,\n contained in board_sizes list for given piece.\n '''\n for board_size in board_sizes:\n board, text_colors = get_board(board_size=board_size)\n plot_board(board, text_colors, piece=piece)", "_____no_output_____" ], [ "special_cases(piece='Pawn', board_sizes=range(4,6))", "_____no_output_____" ] ], [ [ "Закономерность очевидна - всегда присутствует горизонталь (верхняя или нижняя - в зависимости от цвета фигуры), с которой пешка не бьёт ни одну клетку - все поля этой горизонтали $0$-валентны. Их количество равно $n$. \nНа крайних вертикалях расположены $1$-валентные клетки, которых $2(n-1)$ штук.\nВсе остальные поля - $2$-валентны, и расположены они прямоугольником размера $(n-1)\\times(n-2)$.", "_____no_output_____" ], [ "Тогда $$ P(X_{pawn}) = \\frac{n\\cdot 0}{n^{2}(n^{2}-1)} + \\frac{2(n-1)\\cdot 1}{n^{2}(n^{2}-1)} + \\frac{(n-1)(n-2)\\cdot 2}{n^{2}(n^{2}-1)}= \\frac{2(n-1)({\\color{Green}1}+n-{\\color{Green}2})}{n^{2}(n^{2}-1)} = \\frac{2(n-1)^{2}}{n^{2}(n^{2}-1)}. $$ Так как $(n^{2}-1) = (n+1)(n-1)$, $$ P(X_{pawn}) = \\frac{2(n-1)}{n^{2}(n+1)}. $$", "_____no_output_____" ], [ "## Конь", "_____no_output_____" ], [ "### Функция, возвращающая массив валентностей Графа коня", "_____no_output_____" ] ], [ [ "def knight(valencies):\n\n board_size = valencies.shape[0]\n\n if board_size > 3:\n # Four points in each corner are the same for any board size > 3.\n \n # corner cells\n valencies[0, 0] = valencies[0, -1] = \\\n valencies[-1, 0] = valencies[-1, -1] = 2\n\n # cells horizontally/vertically adjacent to the corners\n valencies[0, 1] = valencies[1, 0] = \\\n valencies[0, -2] = valencies[1, -1] = \\\n valencies[-2, 0] = valencies[-1, 1] = \\\n valencies[-2, -1] = valencies[-1, -2] = 3\n\n # cells diagonally adjacent\n valencies[1, 1] = valencies[1, -2] = \\\n valencies[-2, 1] = valencies[-2, -2] = 4\n\n if board_size > 4:\n\n valencies[0, 2:-2] = valencies[2:-2, 0] = \\\n valencies[2:-2, -1] = valencies[-1, 2:-2] = 4\n\n valencies[1, 2:-2] = valencies[2:-2, 1] = \\\n valencies[2:-2, -2] = valencies[-2, 2:-2] = 6\n \n valencies[2:-2, 2:-2] = 8\n\n # Patholigical cases\n elif board_size == 3:\n valencies = 2 * np.ones((board_size, board_size), dtype='uint8')\n valencies[1, 1] = 0\n\n else:\n valencies = np.zeros((board_size, board_size), dtype='uint8')\n\n return valencies", "_____no_output_____" ], [ "special_cases(piece='Knight', board_sizes=[4,5,6])", "_____no_output_____" ] ], [ [ "Количество $2$- и $3$-валентных клеток фиксировано при любом $n\\geq 4$. Первые расположены в углах, а вторые прилегают к ним по вертикали и горизонтали. Стало быть, количество $2$-валентных клеток равно $4$, а $3$-валентных - $8$, вдвое больше. $4$-валентные клетки образуют арифметическую прогрессию с начальным элементом $4$ и шагом $4$ для всех $n\\geq 4$ (при увеличении $n$ на единицу с каждой стороны появляется одна $4$-валентная клетка). Легко видеть, что рост количества $6$-валентных клеток устроен аналогично, однако существуют они только при $n\\geq 5$. Таким образом, $4$-валентных клеток $4(n-3)$, а $6$-валентных клеток - $4(n-4)$ штук. Количество $8$-валентных клеток растёт квадратично, к тому же, они существуют только при $n\\geq 5$. То есть, их количество - $(n-4)^2$. Итого имеем:", "_____no_output_____" ], [ "$$ P(X_{knight}) = \\frac{4\\cdot 2}{n^{2}(n^{2}-1)} + \\frac{8\\cdot 3}{n^{2}(n^{2}-1)} + \\frac{4(n-3)\\cdot 4}{n^{2}(n^{2}-1)} +$$ $$ + \\frac{4(n-4)\\cdot 6}{n^{2}(n^{2}-1)} + \\frac{(n-4)^2\\cdot 8}{n^{2}(n^{2}-1)} = \\frac{32 + 24(n-4) + 16(n-3) + 8(n-4)^{2}}{n^{2}(n^{2}-1)} = $$ \n$$ \\frac{8(4+3(n-4)+2(n-3)+(n-4)^{2})}{n^{2}(n^{2}-1)} = \\frac{8({\\color{Green} 4}+{\\color{Red} {3n}}-{\\color{Green} {12}}+{\\color{Red} {2n}} - {\\color{Green} 6}+ n^{2}-{\\color{Red} {8n}}+{\\color{Green} {16}})}{n^{2}(n^{2}-1)} = $$ \n$$= \\frac{8(n^{2}-3n+2)}{n^{2}(n^{2}-1)} = \\frac{8(n-1)(n-2)}{n^{2}(n^{2}-1)} = \\frac{8(n-2)}{n^{2}(n+1)}. $$", "_____no_output_____" ], [ "## Офицер", "_____no_output_____" ], [ "### Функция, возвращающая массив валентностей Графа офицера (и ферзя)", "_____no_output_____" ], [ "Расположение валентностей для офицера и ферзя практически идентично, за исключением того, что наименьшее значение валентности для ферзя в три раза превышает оное для офицера.", "_____no_output_____" ] ], [ [ "def bishop_or_queen(piece, valencies):\n \n board_size = np.shape(valencies)[0]\n \n if piece == 'Bishop':\n smallest_val = board_size-1\n\n else:\n smallest_val = 3*(board_size-1)\n \n # external square\n valencies[0, :] = valencies[:, 0] = \\\n valencies[:, -1] = valencies[-1, :] = smallest_val\n \n # internal sqares\n for i in range (1, int(board_size/2)+1):\n # top, left\n # right, bottom\n valencies[i, i:-i] = valencies[i:-i, i] = \\\n valencies[i:-i, -(i+1)] = valencies[-(i+1), i:-i] = \\\n smallest_val + 2*i\n \n return valencies", "_____no_output_____" ], [ "special_cases(piece='Bishop', board_sizes=range(4,8))", "_____no_output_____" ] ], [ [ "Видно, что эквивалентные клетки располагаются по периметрам образованных клетками концентрических квадратов. Поскольку при чётных $n$ в центре доски расположены $4$ поля с максимальной валентностью, а при нечётных - одно, случаи чётных и нечётных $n$ представляется удобным рассмотреть раздельно.", "_____no_output_____" ], [ "### Чётные $n$", "_____no_output_____" ], [ "Каково количество различных значений валентности, а также их величина? Наименьшее значение равно $(n-1)$, так как это количество клеток на диагонали минус клетка, на которой стоит сама фигура. Наибольшее значение - $(n-1) + (n-2) = (2n-3)$, так как оно больше наименьшего значения на количество клеток, расположенных на диагонали квадрата со стороной $(n-1)$ минус клетка, на которой стоит сама фигура. ", "_____no_output_____" ], [ "Пусть $s$ - количество шагов размера $2$, которое требуется совершить для перехода от значения $(n-1)$ к значению $(2n-3)$. Тогда\n$$ n-1 + 2s = 2n-3, $$ $$ 2s = {\\color{Red} {2n}} - {\\color{Green} 3} - {\\color{Red} n} + {\\color{Green} 1} = n - 2 \\Rightarrow s = \\frac{n-2}{2}. $$\nТак как $n$ - чётное, $s$ $\\in \\mathbb{Z}$.", "_____no_output_____" ], [ "Однако ввиду того, что *один* шаг совершается между *двумя* разными значениями, количество различных значений валентности на единицу больше количества шагов, требующихся для перехода от минимального до максимального. В таком случае имеем $\\frac{n-2}{2} + 1 = \\frac{n}{2} - {\\color{Green} 1} +{\\color{Green} 1} = \\frac{n}{2}.$ Итого, на доске со стороной $n$ содержится $\\frac{n}{2}$ клеток с различными значениями валентности - $\\frac{n}{2}$ концентрических квадратов.", "_____no_output_____" ], [ "В каком количестве представлено каждое из значений? Количество элементов, расположенных по периметру образованного клетками квадрата со стороной $\\lambda$, равно учетверённой стороне минус четыре угловые клетки, которые оказываются учтёнными дважды. Тогда количество клеток с одноимённым значением валентности равно $4\\lambda-4 = 4(\\lambda-1)$, где $\\lambda$ изменяется с шагом $2$ в пределах от $2$ (центральный квадрат) до $n$ (внешний).", "_____no_output_____" ], [ "При этом от $\\lambda$ зависит не только количество значений валентности, но и их величина - она равна сумме $\\lambda$ и наименьшего из значений валентности, встречающихся на доске. Таким образом, имея наименьшее значение валентности, а также количество концентрических квадратов, нетрудно составить зависимую от $\\lambda$ сумму $P(X^{even}_{bishop}) = \\sum_{}P(H_i) \\cdot P(X|H_i)$. Однако удобнее суммировать по индексу, который изменяется с шагом $1$, потому заменим $k = \\frac{\\lambda}{2}.$ Теперь можно записать:", "_____no_output_____" ], [ "$$ P(X^{even}_{bishop}) = \\sum_{k = 1}^{\\frac{n}{2}} \\frac{4(n+1-2k)\\cdot(n-3+2k)} {n^{2}(n^{2}-1)} = \\frac{4}{n^{2}(n^{2}-1)} \\sum_{k = 1}^{\\frac{n}{2}} n^{2} - {\\color{Red} {3n}} + {\\color{Blue} {2kn}} + {\\color{Red} {n}} - 3 + {\\color{Cyan} {2k}} - {\\color{Blue} {2kn}} + {\\color{Cyan} {6k}} - 4k^{2} = $$ \n\n$$ =\\frac{4}{n^{2}(n^{2}-1)} \\sum_{k = 1}^{\\frac{n}{2}} n^{2} - 2n - 3 + 8k - 4k^{2}. $$ \nВынесем первые три слагаемых за знак суммы, так как они не зависят от $k$, умножив их на $\\frac{n}{2}$ - количество раз, которое они встречаются в сумме: \n\n$$ P(X^{even}_{bishop}) = \\frac{4}{n^{2}(n^{2}-1)}\\ [\\frac{n}{2}(n^{2} - 2n - 3) + \\sum_{k = 1}^{\\frac{n}{2}}8k - 4k^{2}] $$", "_____no_output_____" ], [ "Рассмотрим отдельно выражение под знаком суммы.", "_____no_output_____" ], [ "$$ \\sum_{k = 1}^{\\frac{n}{2}}8k - 4k^{2} = 8\\sum_{k = 1}^{\\frac{n}{2}}k - 4\\sum_{k = 1}^{\\frac{n}{2}} k^{2}. $$", "_____no_output_____" ], [ "Обозначим $ S_1 = 8\\sum_{k = 1}^{\\frac{n}{2}}k$, $ S_2 = 4\\sum_{k = 1}^{\\frac{n}{2}} k^{2}. $", "_____no_output_____" ], [ "$S_1$ - это умноженная на $8$ сумма первых $\\frac{n}{2}$ натуральных чисел, которая есть сумма первых $\\frac{n}{2}$ членов арифметической прогрессии, поэтому ", "_____no_output_____" ], [ "$$ S_1 = 8\\frac{\\frac{n}{2}(\\frac{n}{2}+1)}{2} = 4\\frac{n}{2}(\\frac{n}{2}+1) = 2n(\\frac{n}{2}+1) = \\frac{2n^2}{2}+2n = n^2 + 2n = n(n+2). $$", "_____no_output_____" ], [ "$S_2$ - это умноженная на 4 сумма квадратов первых $\\frac{n}{2}$ натуральных чисел, поэтому", "_____no_output_____" ], [ "$$ S_2 = 4\\frac{\\frac{n}{2}(\\frac{n}{2}+1)(2\\frac{n}{2}+1)}{6} = \\frac{n(n+2)(n+1)}{6}. $$", "_____no_output_____" ], [ "$$ S_1 - S_2 = n(n+2) - \\frac{n(n+2)(n+1)}{6} = n(n+2) (1 - \\frac{(n+1)}{6}) = $$ $$ = \\frac{n(n+2)({\\color{Green} 6}-n-{\\color{Green} 1})}{6} = \\frac{n(n+2)(-n + 5)}{6} = -\\frac{n(n+2)(n-5)}{6}.$$ ", "_____no_output_____" ], [ "Тогда", "_____no_output_____" ], [ "$$ P(X^{even}_{bishop}) = \\frac{4}{n^{2}(n^{2}-1)}\\ [\\frac{n}{2}(n^{2} - 2n - 3) - \\frac{n(n+2)(n-5)}{6} ] = $$ $$ = \\frac{4}{n^{2}(n^{2}-1)}\\ [\\frac{n(3n^{2} - 6n - 9)}{6} - \\frac{n(n+2)(n-5)}{6} ] = $$ $$ = \\frac{4n}{6n^{2}(n^{2}-1)}({\\color{Orange} {3n^{2}}} - {\\color{Red} {6n}} - {\\color{Green} 9} - {\\color{Orange} {n^2}} + {\\color{Red} {5n}} - {\\color{Red} {2n}} + {\\color{Green} {10}}) = $$ $$ =\\frac{2}{3n(n^{2}-1)}(2n^2 - 3n + 1) = \\frac{2(2n-1)(n-1)}{3n(n^{2}-1)} = \\frac{2(2n-1)}{3n(n+1)}. $$", "_____no_output_____" ], [ "### Нечётные $n$", "_____no_output_____" ], [ "Каково количество различных значений валентности? Наименьшее значение равно $(n-1)$ из тех же рассуждений, что и для чётных $n$. Наибольшее значение, очевидно, равно удвоенному наименьшему - $(n-1) + (n-1) = 2(n-1)$. ", "_____no_output_____" ], [ "Пусть $s$ - количество шагов размера $2$, которое требуется совершить для перехода от значения $(n-1)$ к значению $2(n-1)$. Тогда\n$$n-1 + 2s = 2n-2,$$ $$2s = {\\color{Red} {2n}} - {\\color{Green} 2} - {\\color{Red} n} + {\\color{Green} 1} = n - 1 \\Rightarrow s = \\frac{n-1}{2}.$$\nТак как $n$ - нечётное, $s$ $\\in \\mathbb{Z}$. Итого имеем $\\frac{n-1}{2} + 1 = \\frac{n}{2} - {\\color{Green} {\\frac{1}{2}}} +{\\color{Green} 1} = \\frac{n}{2} + \\frac{1}{2} = \\frac{n+1}{2}$ клеток с различными значениями валентности.", "_____no_output_____" ], [ "В каком количестве представлено каждое из значений? Рассуждения для чётных и нечётных $n$ идентичны, за исключением того, что выражение $4(\\lambda-1)$ равно нулю при $\\lambda = 1$ (в центральной клетке доски). По этой причине слагаемое $P(H_{\\frac{n+1}{2}}) \\cdot P(X|H_{\\frac{n+1}{2}})$ должно быть вынесено за знак общей суммы, а индекс суммирования будет принимать на единицу меньше значений: $\\frac{n+1}{2} - 1 = \\frac{n}{2} + \\frac{1}{2} - 1 = \\frac{n}{2} + {\\color{Green} {\\frac{1}{2}}} - {\\color{Green} 1} = \\frac{n}{2} - \\frac{1}{2} = \\frac{n-1}{2}.$", "_____no_output_____" ], [ "Тогда можно записать:", "_____no_output_____" ], [ "$$ P(X^{odd}_{bishop}) = \\frac{1\\cdot 2(n-1)}{n^{2}(n^{2}-1)} + \\sum_{k = 1}^{\\frac{n-1}{2}} \\frac{4(n+1-2k)\\cdot(n-3+2k)} {n^{2}(n^{2}-1)}. $$", "_____no_output_____" ], [ "Легко видеть, что выражение под знаком суммы отличается от $P(X^{even}_{bishop})$ только верхней границей суммирования. Следовательно, аналогично предыдущим выкладкам можно обозначить: $ S_1 = 8\\sum_{k = 1}^{\\frac{n-1}{2}}k$, $ S_2 = 4\\sum_{k = 1}^{\\frac{n-1}{2}} k^{2}. $", "_____no_output_____" ], [ "$$ S_1 = 8\\frac{\\frac{n-1}{2}(\\frac{n-1}{2}+1)}{2} = 4\\frac{n-1}{2}(\\frac{n+1}{2}) = (n-1)(n+1). $$", "_____no_output_____" ], [ "$$ S_2 = 4\\frac{\\frac{n-1}{2}(\\frac{n-1}{2}+1)(2\\frac{n-1}{2}+1)}{6} = 4\\frac{\\frac{n-1}{2}(\\frac{n-1}{2}+1)(2\\frac{n-1}{2}+1)}{6} = \\frac{(n-1)(\\frac{n+1}{2})n}{3} = \\frac{(n-1)(n+1)n}{6}. $$", "_____no_output_____" ], [ "$$ S_1 - S_2 = (n-1)(n+1) - \\frac{(n-1)(n+1)n}{6} = (n-1)(n+1)(1 - \\frac{n}{6}) = \\frac{(n-1)(n+1)(6 - n)}{6} = -\\frac{(n-1)(n+1)(n-6)}{6}. $$", "_____no_output_____" ], [ "Тогда", "_____no_output_____" ], [ "$$ P(X^{odd}_{bishop}) = \\frac{2(n-1)}{n^{2}(n^{2}-1)} + \\frac{4}{n^{2}(n^{2}-1)}\\ [\\frac{n-1}{2}(n^{2} - 2n - 3) -\\frac{(n-1)(n+1)(n-6)}{6}] = $$ $$ =\\frac{2}{n^{2}(n+1)} + \\frac{4(n-1)}{n^{2}(n^{2}-1)} [\\frac{3n^2 - 6n - 9}{6} -\\frac{(n+1)(n-6)}{6}] = $$ $$ =\\frac{2}{n^{2}(n+1)} + \\frac{4}{6n^{2}(n+1)}({\\color{Orange} {3n^2}} - {\\color{Red} {6n}} - {\\color{Green} 9} - {\\color{Orange} {n^2}} + {\\color{Red} {6n}} - {\\color{Red} n} + {\\color{Green} 6}) = $$ $$ =\\frac{2}{n^{2}(n+1)} + \\frac{4}{6n^{2}(n+1)}(2n^2 - n - 3) = \\frac{{\\color{Green} {12}} + 8n^2 - 4n - {\\color{Green} {12}}}{6n^{2}(n+1)} = \\frac{4n(2n-1)}{6n^{2}(n+1)} = \\frac{2(2n-1)}{3n(n+1)}. $$", "_____no_output_____" ], [ "Как видно, чётность доски не влияет на рассматриваемую вероятность: $P(X^{even}_{bishop}) = P(X^{odd}_{bishop}) = P(X_{bishop}) = \\frac{2(2n-1)}{3n(n+1)}$.", "_____no_output_____" ], [ "## Ладья", "_____no_output_____" ], [ "### Функция, возвращающая массив валентностей ладейного Графа", "_____no_output_____" ] ], [ [ "def rook(valencies):\n \n board_size = np.shape(valencies)[0]\n \n x, y = np.meshgrid(range(board_size), range(board_size))\n \n for i, j in zip(x.flatten(), y.flatten()):\n valencies[i, j] = 2*(board_size-1)\n \n return valencies ", "_____no_output_____" ], [ "special_cases(piece='Rook', board_sizes=range(4,6))", "_____no_output_____" ] ], [ [ "Известная особенность ладьи - независимо от расположения на доске, она всегда контролирует постоянное количество полей, а именно $2(n-1)$ - это сумма полей по горизонтали и вертикали минус поле, на котором стоит сама ладья.\n$$P(X_{rook}) = \\frac{n^{2}\\cdot 2(n-1)}{n^{2}(n^{2}-1)} = \\frac{2}{(n+1)}.$$", "_____no_output_____" ], [ "## Ферзь", "_____no_output_____" ] ], [ [ "special_cases(piece='Queen', board_sizes=range(4,8))", "_____no_output_____" ] ], [ [ "Поскольку ферзь сочетает в себе возможности офицера и ладьи, выражение для него может быть получено как сумма выражений для этих фигур:", "_____no_output_____" ], [ "$$ P(X_{queen}) = \\frac{2(2n-1)}{3n(n+1)} + \\frac{2}{n+1} = \\frac{2(2n-1) + 6n}{3n(n+1)} = \\frac{{\\color{Red} {4n}} - 2 + {\\color{Red} {6n}}}{3n(n+1)} = \\frac{10n - 2}{3n(n+1)} = \\frac{2(5n-1)}{3n(n+1)}. $$", "_____no_output_____" ], [ "## Король", "_____no_output_____" ], [ "### Функция, возвращающая массив валентностей Графа короля", "_____no_output_____" ] ], [ [ "def king(valencies):\n \n # corners : top left = top right = \\\n # bottom left = bottom right\n valencies[0, 0] = valencies[0, -1] = \\\n valencies[-1, 0] = valencies[-1, -1] = 3\n \n # edges : top, left, right, bottom\n valencies[0, 1:-1] = valencies[1:-1, 0] = \\\n valencies[1:-1, -1] = valencies[-1, 1:-1] = 5\n \n # center\n valencies[1:-1, 1:-1] = 8\n \n return valencies", "_____no_output_____" ], [ "special_cases(piece='King', board_sizes=range(4,6))", "_____no_output_____" ] ], [ [ "Видно, что края доски, за исключением $3$-валентных углов, $5$-валентны, а всё оставшееся пространство $8$-валентно. Ввиду того, что краёв $4$, а $5$-валентных клеток на одном краю $(n-2)$ штук, имеем: \n$$ P(X_{king}) = \\frac{4\\cdot 3}{n^{2}(n^{2}-1)} +\\frac{4(n-2)\\cdot 5}{n^{2}(n^{2}-1)} +\\frac{(n-2)^2\\cdot 8}{n^{2}(n^{2}-1)} = \\frac{12 + 20(n-2) + 8(n-2)^2}{n^{2}(n^{2}-1)} = $$\n$$ = \\frac{4(3 + 5(n-2)+2(n-2)^2)}{n^{2}(n^{2}-1)} = \\frac{4(3 + 5n-10+2(n^2 - 4n + 4))}{n^{2}(n^{2}-1)} = \\frac{4({\\color{Green} 3} + {\\color{Red} {5n}}-{\\color{Green} {10}}+2n^2 - {\\color{Red} {8n}} + {\\color{Green} {8}} )}{n^{2}(n^{2}-1)} = $$ \n$$ =\\frac{4(2n^2 - 3n + 1)}{n^{2}(n^{2}-1)} = \\frac{4(2n-1)(n-1)}{n^{2}(n^{2}-1)} = \\frac{4(2n-1)}{n^{2}(n+1)}. $$", "_____no_output_____" ], [ "### Функция, возвращающая значение $P(X_{piece})$", "_____no_output_____" ] ], [ [ "def get_probabilities(piece, n):\n \n # NOTE: Results can be incorrect for large n because of dividing by\n # the huge denominator!\n if piece == 'Pawn':\n return 2*(n-1)/((n**2)*(n+1)) \n \n elif piece == 'Knight':\n return 8*(n-2)/((n**2)*(n+1)) \n\n elif piece == 'Bishop':\n return 2*(2*n-1)/(3*n*(n+1)) \n \n elif piece == 'Rook':\n return 2/(n+1) \n\n elif piece == 'Queen':\n return 2*(5*n-1)/(3*n*(n+1)) \n\n elif piece == 'King':\n return 4*(2*n-1)/(n**2*(n+1))\n", "_____no_output_____" ] ], [ [ "Для проверки аналитических результатов используем метод вычисления вероятности «в лоб» - напрямую из массива валентностей.", "_____no_output_____" ] ], [ [ "def straightforward_prob(piece, board_size):\n # Get probability directly from the board of valencies\n board, _ = get_board(board_size)\n val_board = get_valencies(piece, board)\n \n unique, counts = np.unique(val_board, return_counts=True)\n prob = np.dot(unique, counts)/((board_size)**2 * (board_size**2 - 1))\n \n return prob ", "_____no_output_____" ] ], [ [ "График, отображающий зависимость вероятности от размера доски, представлен как функция действительного переменного в целях наглядности. ", "_____no_output_____" ] ], [ [ "start = 2\nend = 16\nstep = 0.02\nx = np.arange(start, end)\nnames_list = ['Pawn', 'Knight', 'Bishop', 'Rook', 'Queen', 'King']\n\n# Check if analytical results match straightforward calculations\nfor name in names_list:\n for board_size in x:\n y = get_probabilities(name, board_size)\n if not y == straightforward_prob(name, board_size):\n print('Mistake in equation for %s' % name)\nprint('Analytical results approved')\n\n# Let's expand the range from Z to R for the sake of visual clarity\nx = np.arange(start, end, step)\nfig, ax = plt.subplots(figsize=(10, 8)) \n\nfor name in names_list:\n y = get_probabilities(name, x) \n plt.plot(x, y, label=name, linewidth=3.0)\n \nlegend = plt.legend(loc='upper right')\n\nfor label in legend.get_lines():\n label.set_linewidth(3) \n\nfor label in legend.get_texts(): \n label.set_fontsize(26)\n\nplt.xlabel(\"Size of a board\", fontsize=20) \nplt.ylabel(\"Probability\", fontsize=20) \nplt.show() ", "Analytical results approved\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Collaborative Filtering on Google Analytics Data\n", "_____no_output_____" ], [ "### Learning objectives\n1. Prepare the user-item matrix and use it with WALS.\n2. Train a `WALSMatrixFactorization` within TensorFlow locally and on AI Platform.\n3. Visualize the embedding vectors with principal components analysis.\n", "_____no_output_____" ], [ "## Overview\nThis notebook demonstrates how to implement a WALS matrix refactorization approach to do collaborative filtering.\n\nEach learning objective will correspond to a __#TODO__ in the [student lab notebook](../labs/wals.ipynb) -- try to complete that notebook first before reviewing this solution notebook.\n", "_____no_output_____" ] ], [ [ "import os\nPROJECT = \"cloud-training-demos\" # REPLACE WITH YOUR PROJECT ID\nBUCKET = \"cloud-training-demos-ml\" # REPLACE WITH YOUR BUCKET NAME\nREGION = \"us-central1\" # REPLACE WITH YOUR BUCKET REGION e.g. us-central1\n\n# Do not change these\nos.environ[\"PROJECT\"] = PROJECT\nos.environ[\"BUCKET\"] = BUCKET\nos.environ[\"REGION\"] = REGION\nos.environ[\"TFVERSION\"] = \"1.15\"", "_____no_output_____" ], [ "%%bash\ngcloud config set project $PROJECT\ngcloud config set compute/region $REGION", "Updated property [core/project].\nUpdated property [compute/region].\n" ], [ "import tensorflow as tf\nprint(tf.__version__)", "1.15.5\n" ] ], [ [ "## Create raw dataset\n<p>\nFor collaborative filtering, you don't need to know anything about either the users or the content. Essentially, all you need to know is userId, itemId, and rating that the particular user gave the particular item.\n<p>\nIn this case, you are working with newspaper articles. The company doesn't ask their users to rate the articles. However, you can use the time-spent on the page as a proxy for rating.\n<p>\nNormally, you would also add a time filter to this (\"latest 7 days\"), but your dataset is itself limited to a few days.", "_____no_output_____" ] ], [ [ "from google.cloud import bigquery\nbq = bigquery.Client(project = PROJECT)\n\nsql = \"\"\"\nWITH CTE_visitor_page_content AS (\n SELECT\n # Schema: https://support.google.com/analytics/answer/3437719?hl=en\n # For a completely unique visit-session ID, you combine combination of fullVisitorId and visitNumber:\n CONCAT(fullVisitorID,'-',CAST(visitNumber AS STRING)) AS visitorId,\n (SELECT MAX(IF(index=10, value, NULL)) FROM UNNEST(hits.customDimensions)) AS latestContentId, \n (LEAD(hits.time, 1) OVER (PARTITION BY fullVisitorId ORDER BY hits.time ASC) - hits.time) AS session_duration \n FROM\n `cloud-training-demos.GA360_test.ga_sessions_sample`, \n UNNEST(hits) AS hits\n WHERE \n # only include hits on pages\n hits.type = \"PAGE\"\nGROUP BY \n fullVisitorId,\n visitNumber,\n latestContentId,\n hits.time )\n-- Aggregate web stats\nSELECT \n visitorId,\n latestContentId as contentId,\n SUM(session_duration) AS session_duration\nFROM\n CTE_visitor_page_content\nWHERE\n latestContentId IS NOT NULL \nGROUP BY\n visitorId, \n latestContentId\nHAVING \n session_duration > 0\n\"\"\"\n\ndf = bq.query(sql).to_dataframe()\ndf.head()", "_____no_output_____" ], [ "stats = df.describe()\nstats", "_____no_output_____" ], [ "df[[\"session_duration\"]].plot(kind=\"hist\", logy=True, bins=100, figsize=[8,5])", "_____no_output_____" ], [ "# The rating is the session_duration scaled to be in the range 0-1. This will help with training.\nmedian = stats.loc[\"50%\", \"session_duration\"]\ndf[\"rating\"] = 0.3 * df[\"session_duration\"] / median\ndf.loc[df[\"rating\"] > 1, \"rating\"] = 1\ndf[[\"rating\"]].plot(kind=\"hist\", logy=True, bins=100, figsize=[8,5])", "_____no_output_____" ], [ "del df[\"session_duration\"]", "_____no_output_____" ], [ "%%bash\nrm -rf data\nmkdir data", "_____no_output_____" ], [ "# TODO 1: Write object to a comma-separated values (csv) file.\ndf.to_csv(path_or_buf = \"data/collab_raw.csv\", index = False, header = False)", "_____no_output_____" ], [ "!head data/collab_raw.csv", "1012012094517511217-1,299949290,1.0\n1614877438160481626-349,29455408,1.0\n1815317450115839569-235,299807267,0.8569622728727375\n1950053456137072413-10,243386345,0.5474309284215766\n2179122255490468209-180,299925086,0.03222095793478647\n2477455524323251766-1,299936493,0.3514756604914127\n277149937495438413-384,299777707,0.1444742737261316\n2912361176772133138-147,299796840,1.0\n3868650583032521586-577,299814183,0.5796732212068414\n4086929260756256102-380,299946732,1.0\n" ] ], [ [ "## Create dataset for WALS\n<p>\nThe raw dataset (above) won't work for WALS:\n<ol>\n<li> The userId and itemId have to be 0,1,2 ... so you need to create a mapping from visitorId (in the raw data) to userId and contentId (in the raw data) to itemId.\n<li> You will need to save the above mapping to a file because at prediction time, you'll need to know how to map the contentId in the table above to the itemId.\n<li> You'll need two files: a \"rows\" dataset where all the items for a particular user are listed; and a \"columns\" dataset where all the users for a particular item are listed.\n</ol>\n\n<p>\n\n### Mapping", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport numpy as np\ndef create_mapping(values, filename):\n with open(filename, 'w') as ofp:\n value_to_id = {value:idx for idx, value in enumerate(values.unique())}\n for value, idx in value_to_id.items():\n ofp.write(\"{},{}\\n\".format(value, idx))\n return value_to_id\n\ndf = pd.read_csv(filepath_or_buffer = \"data/collab_raw.csv\",\n header = None,\n names = [\"visitorId\", \"contentId\", \"rating\"],\n dtype = {\"visitorId\": str, \"contentId\": str, \"rating\": np.float})\ndf.to_csv(path_or_buf = \"data/collab_raw.csv\", index = False, header = False)\nuser_mapping = create_mapping(df[\"visitorId\"], \"data/users.csv\")\nitem_mapping = create_mapping(df[\"contentId\"], \"data/items.csv\")", "_____no_output_____" ], [ "!head -3 data/*.csv", "==> data/collab_raw.csv <==\n1012012094517511217-1,299949290,1.0\n1614877438160481626-349,29455408,1.0\n1815317450115839569-235,299807267,0.8569622728727375\n\n==> data/items.csv <==\n299949290,0\n29455408,1\n299807267,2\n\n==> data/users.csv <==\n1012012094517511217-1,0\n1614877438160481626-349,1\n1815317450115839569-235,2\n" ], [ "df[\"userId\"] = df[\"visitorId\"].map(user_mapping.get)\ndf[\"itemId\"] = df[\"contentId\"].map(item_mapping.get)", "_____no_output_____" ], [ "mapped_df = df[[\"userId\", \"itemId\", \"rating\"]]\nmapped_df.to_csv(path_or_buf = \"data/collab_mapped.csv\", index = False, header = False)\nmapped_df.head()", "_____no_output_____" ] ], [ [ "### Creating rows and columns datasets", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport numpy as np\nmapped_df = pd.read_csv(filepath_or_buffer = \"data/collab_mapped.csv\", header = None, names = [\"userId\", \"itemId\", \"rating\"])\nmapped_df.head()", "_____no_output_____" ], [ "NITEMS = np.max(mapped_df[\"itemId\"]) + 1\nNUSERS = np.max(mapped_df[\"userId\"]) + 1\nmapped_df[\"rating\"] = np.round(mapped_df[\"rating\"].values, 2)\nprint(\"{} items, {} users, {} interactions\".format( NITEMS, NUSERS, len(mapped_df) ))", "5662 items, 120983 users, 284343 interactions\n" ], [ "grouped_by_items = mapped_df.groupby(\"itemId\")\niter = 0\nfor item, grouped in grouped_by_items:\n print(item, grouped[\"userId\"].values, grouped[\"rating\"].values)\n iter = iter + 1\n if iter > 5:\n break", "0 [ 0 357 561 608 746 1119 1159 1207 1315 1348 1370 1396\n 1940 2035 2143 2254 3169 3313 176 3692 611 4003 4528 1432\n 4625 4862 4879 5110 5324 5409 5506 5693 5732 5743 5838 491\n 7167 7349 7908 7919 8064 8217 5758 8764 8829 9050 9140 9311\n 9417 9820 10368 10418 10809 10822 10911 11351 11761 9110 11982 3345\n 12292 12599 12841 12863 13342 13758 13986 14038 14153 14357 15069 15074\n 15106 15346 15424 15519 13030 16265 16363 5322 16416 16560 16749 17042\n 17077 17403 17442 6664 17575 17674 17735 17916 17990 18152 18181 13354\n 18383 18532 16206 13652 18643 18905 11473 19213 19655 19685 19699 17437\n 19817 3948 20195 15545 20285 20733 20792 16549 21222 5923 21627 21796\n 12084 14759 22053 22116 22691 1287 22717 20511 23252 23446 23479 5683\n 23796 17283 24177 24348 810 10004 24800 25200 13567 1899 8200 25455\n 25570 25629 25806 25822 25870 25904 11926 26330 26598 26610 26751 26914\n 27307 28187 6666 579 28556 28604 28708 28731 28756 24762 28916 28966\n 29452 29521 11392 11503 29771 29829 29935 12012 30161 30496 3827 26662\n 30686 30746 17931 30809 13231 30980 20602 31324 31418 31621 31898 3034\n 31997 389 32253 32413 32502 12924 32719 32744 32817 32890 32929 5062\n 29483 33477 33528 14164 33617 33642 33703 33771 33834 34045 34066 34101\n 3455 28351 9561 34206 34522 4378 34789 34857 34994 35004 2160 35210\n 35447 35588 35605 35725 35815 35818 36006 36111 26557 24649 36384 26845\n 4605 22817 22916 37058 37093 37389 19517 9461 37818 6890 22314 37945\n 38436 33298 21167 38896 25831 19582 39375 39554 28739 32642 39832 40055\n 4750 40269 31276 40332 21443 11597 350 41247 41345 41441 918 4525\n 18200 25024 15950 40285 8029 42042 2197 11288 8735 30551 890 32677\n 43227 34610 43447 27111 1548 31044 1807 43717 21284 44031 11636 2802\n 44168 44245 21948 44417 12241 6776 45088 45172 38633 45347 2248 45488\n 45576 8867 45720 37552 35866 45944 46118 46227 41742 46690 46730 1973\n 46791 46843 46892 46937 5483 47006 29697 14330 47309 24170 47509 47611\n 47882 7307 4978 38715 43861 37051 48606 19146 35573 3092 30196 48951\n 49159 49214 4064 39936 41673 49432 49461 49488 10504 49776 11221 49973\n 50009 5664 6297 42753 22220 34315 15468 50756 38125 38223 50883 50957\n 25159 51131 51187 18763 5550 51564 17084 51793 39490 51911 44702 52123\n 52484 8354 8701 11873 46 53227 53297 15206 18026 15900 53759 53760\n 7810 53840 10923 53895 20950 5332 43946 54472 54477 54496 6715 4022\n 7296 54877 54980 30992 27101 55045 55085 41871 55100 48350 2392 8750\n 55727 55746 6746 55976 12697 12787 56230 56266 53648 31410 42156 56638\n 23987 57145 57182 41351 24499 4302 7263 18016 55019 40434 2871 28104\n 40941] [1. 0.19 0.93 0.55 0.4 0.54 0. 0.09 0.57 1. 0.36 0.43 0.37 0.44\n 0.57 0.03 0.08 0.02 0.16 0.36 0.07 0.72 0.87 0.04 0.64 0.69 0.05 0.4\n 0.14 0.03 0.06 0.36 0.6 0.03 0.07 0.01 0.6 0.56 0.36 0.75 0.36 1.\n 1. 0.16 0.58 1. 0.98 0.37 1. 0.92 0.45 0.49 0.47 1. 1. 0.19\n 0.01 0.09 0.07 0.27 0.34 0.21 0.01 0.01 0.91 0.57 0.28 0.92 0.02 0.65\n 0.25 0.87 0.46 1. 0. 1. 0.37 1. 0.53 1. 1. 0.42 0.1 1.\n 0.47 1. 1. 0.41 0.04 0.44 1. 0.34 0.81 0.56 0.32 0.04 0.25 0.03\n 0.73 1. 0.03 0.21 0.05 0.11 0.26 1. 0.67 1. 0.4 0.69 0.19 0.1\n 0.14 0.98 0.46 0.46 0.11 0.94 0.46 0.92 0.1 1. 0.39 1. 0.67 0.23\n 0.76 1. 1. 0.68 0.68 0.25 0.41 0.79 0.81 0.05 0.72 0.04 0.07 0.34\n 1. 0.61 0.32 1. 0.37 1. 0.81 0.41 0.78 0.39 0.02 0.27 0.91 1.\n 0.21 1. 0.19 1. 0.33 0.66 0.56 0.04 0.43 0.78 0.47 0.93 0.52 0.15\n 0.61 0.2 1. 0.45 0.03 0.48 0.09 0.44 0.14 0.14 0.01 0.05 0.01 0.52\n 0.36 0.07 1. 0.13 0.57 1. 1. 1. 0.74 0.63 0.02 0.03 0.17 0.63\n 0.54 0.02 0.72 0.07 0.43 1. 1. 0.01 0.62 0.05 1. 0.03 0.26 0.48\n 0.5 1. 0.46 0.3 0.14 0.57 0.51 0.75 0.8 0.01 0.02 0.6 1. 1.\n 0.08 0.48 1. 0.18 0.91 0.54 0.3 0.21 0.52 0.03 0.03 1. 0.06 0.46\n 0.05 0.47 0.42 0.32 0.13 0.25 0.95 0.04 0.92 1. 0.5 0.45 0.32 0.99\n 0.6 0.78 0.17 0.32 0.24 0.77 0.46 0.66 0.15 0.78 0.27 0.31 0.36 0.25\n 0.76 0.04 0.43 1. 0.43 0.26 0.63 0.78 0.62 0.48 0.48 1. 0.08 0.01\n 0.48 0.41 0.17 0.49 0.45 0.69 0.14 0.76 1. 1. 0.58 0.02 0.52 1.\n 0.87 0.03 0.08 0.73 0.03 0. 0.78 0.28 0.27 0.99 0.62 0.25 0.37 0.95\n 0.37 0.96 0.45 0.16 0.08 1. 0.24 0. 0.03 0.04 0.5 0.21 0.31 0.09\n 0.85 0.03 0.4 0.07 0.47 0.41 0.62 0.01 0.89 0.07 0.29 0.04 0.92 1.\n 0.16 0.04 0.6 0.27 0.57 0.15 0.56 0.93 0.28 1. 0.69 0.61 0.32 0.01\n 0.49 1. 0.07 0.72 0.28 1. 0.68 0.04 0.26 0.28 0.09 0.03 1. 0.78\n 1. 0.4 0.55 0.36 0.04 0.62 1. 0.5 0.12 1. 0.33 1. 0.46 0.05\n 0.09 0.67 0.72 0.49 1. 0.47 0.57 0.38 0.91 1. 0.01 0.41 0.04 1.\n 0.19 0.06 0.68 0.77 0.11 0.31 0.15 0.33 0.47 0.91 0.24 0.11 0.4 1.\n 0.64 0.11 1. 0.08 1. 0.6 0.04 0.34 0.38 0.26 1. 0.64 0.87 0.24\n 0.45 0.01 1. 1. 0.45 0.09 1. 0.43 0.03 0.47 0.8 0.41 0.44 0.14\n 0.49 0.31 0.54 0.02 0.09 0.68 0.12 0.14 0.63 0.3 0.76]\n1 [ 1 2468] [1. 0.7]\n2 [ 2 1102 1529 2209 2413 2581 3155 3185 3365 4460 4792 5042\n 5789 6980 6992 4111 7403 8052 8308 8340 8556 8895 8900 9868\n 10187 10376 7594 10775 8029 11270 11376 11448 11476 11694 12310 506\n 12741 13139 13329 7864 5291 14267 14268 252 14961 15235 7685 10716\n 17004 11878 17652 17680 18180 18217 18310 3300 19828 20221 20732 15562\n 1973 23535 14400 23824 23971 24335 3162 26143 26206 26495 26980 27012\n 2311 21946 28358 29211 29313 23310 11251 29759 8261 32172 32538 32573\n 32609 32652 33063 19200 33716 34004 34027 26366 34225 7478 34974 1822\n 29483 36061 36252 36515 37262 31746 37485 37525 38397 38712 39130 39311\n 39542 17502 40027 5166 37229 21481 11594 16836 21615 40891 9510 28679\n 41525 41646 32836 31141 41886 13820 21732 41299 10152 7790 43557 43733\n 45115 42408 14421 9141 9847 36258 46677 29794 25806 33827 21989 47656\n 34544 47983 48065 48113 36663 8812 48672 21872 4671 16359 49926 11663\n 3563 32327 50606 1237 50914 18245 51523 25907 12246 51978 935 52065\n 52407 48171 31205 19055 39118 2893 11765 576 7143 53514 43349 53634\n 39075 24216 54609 14269 2859 55690 56129 13214 2195 8700 40892 3061\n 57183 34328 57368 57392 57453 57577 24940 15854 51041 58076 44054] [0.86 0.41 0.47 0.05 0.51 0.66 0.38 0.35 0.34 1. 0.63 0.03 0.12 1.\n 0.08 0.14 0.09 0.4 0.29 0.12 0.58 0.36 0.29 1. 0.77 0.6 0.13 0.16\n 0.25 0.57 0.28 1. 0.67 0.36 0.65 1. 1. 0.66 1. 1. 0.83 0.02\n 0.04 0.53 0.35 0.5 1. 0.31 0.53 0.12 1. 1. 0. 0.21 0.37 0.07\n 0.26 0.05 1. 0.27 0.36 0.45 0.62 0.4 0.17 0.06 0.01 0.17 1. 0.15\n 0.06 0.49 0.02 0.08 0.87 0.7 1. 0.12 0.28 0.27 0.37 1. 1. 1.\n 1. 0.65 0.1 1. 1. 1. 0.09 0.26 0.34 0.64 0.63 0.23 0.41 0.17\n 0.02 0.43 0.15 0.23 0.3 0.77 1. 0.03 0.3 0.29 1. 0.03 0.03 0.02\n 0.63 0.02 0.47 0.88 0.17 1. 0.01 0.55 1. 1. 0.3 1. 0.18 0.08\n 1. 0.14 1. 0.26 1. 1. 0.3 0.12 0.05 0.46 0.44 0.72 0.01 0.09\n 0.12 0.08 0.96 0.29 1. 0.23 1. 0.28 0.13 0.7 1. 0.34 0.41 0.23\n 1. 0.17 0.46 0.1 0.33 0.21 1. 0.22 0.31 0.05 0.94 0.85 0.27 0.4\n 1. 0.63 1. 0.86 0.19 0.05 0.42 0.1 0.39 1. 1. 1. 0.26 0.11\n 0.25 0.66 0.85 0.22 1. 0.57 0.42 0.36 1. 0.01 1. 0.03 1. 0.21\n 0.5 1. 0.02 0.27 0.97 0.22 0.23]\n3 [ 3 229 900 1536 1981 2145 3838 4953 5330 5621 5691 5812\n 2819 5912 4244 8307 8393 9131 9233 9502 10069 1583 11109 2172\n 11343 11523 11678 5900 11742 12195 12458 13794 14493 11849 15437 15626\n 16240 16306 17400 17487 943 19053 19166 19696 20219 20813 21088 14139\n 21438 21452 2769 14680 22111 1871 23358 18875 23458 23747 277 12870\n 15755 25332 25413 25429 23586 25680 25784 25950 26765 26787 25493 27777\n 27784 8781 28011 17464 17581 3867 29041 29193 30621 30648 31975 9890\n 15607 10167 34736 20722 35220 35542 9065 6254 35804 24425 1571 5514\n 37175 37571 38077 38505 38513 38626 38696 2380 38866 35728 26277 26306\n 39862 20435 40303 40561 9149 9816 20673 42241 23790 43064 28689 26975\n 17580 4789 29170 33382 45426 45521 19523 534 41264 46245 30931 40558\n 8763 45661 39787 17924 22876 48134 2196 48646 46138 4013 10587 49717\n 50057 50196 50482 50773 51551 9846 52286 52329 52691 11639 2848 52949\n 45879 39435 53681 16032 54144 54364 3390 54672 9912 22638 42022 8134\n 23284 21698 6364 14871 1254 34824 25050 56524 56637 14496 57294 7553\n 57764 16540 58108] [0.55 0.75 0.2 0. 1. 0.45 0.08 1. 0.08 0.2 0.04 0.49 0.1 0.34\n 0.27 0.35 0.09 0.89 0.13 0.16 0.18 0.38 0.46 0.6 0.28 0.57 0.8 0.12\n 0.27 0.75 0.53 0.68 0.06 0.29 1. 0.15 0.68 0.63 0.44 1. 0.33 0.16\n 0.29 0.5 0.02 0.2 0.04 0.38 0.22 0.02 0.67 0.41 0.28 1. 1. 0.28\n 0.35 1. 0.61 0.54 0.15 0.24 0.5 0.37 0.38 0.4 1. 0.39 0.12 0.6\n 0.38 0.03 0.08 0.89 1. 0.37 0.23 0.39 0.02 0.18 0.33 0.24 0.85 0.73\n 1. 0.7 0.33 0.37 1. 0.6 1. 0.59 1. 1. 0.41 0.16 0.1 0.35\n 0.67 0.51 0.28 0.56 0.5 0.39 0.02 0.56 0.53 0.63 0.59 0.16 0.05 0.19\n 0.2 1. 0.32 0.7 0.78 0.12 0.05 0.43 0.51 0.37 0.5 0.14 1. 0.27\n 0.2 1. 0.92 0.07 0.04 0.41 0.03 0.55 0.31 0.07 0.47 1. 0.07 0.34\n 0.02 0.76 0.6 0.43 0.48 0.31 0.41 0.13 0.7 0.49 0.03 0.09 0.08 0.43\n 0.34 0.26 0.15 1. 0.13 0.41 1. 0.4 0.14 0.25 0.62 0.71 0.6 0.32\n 0.72 0.24 0.33 0.45 0.48 0.75 0.1 0.38 0.3 0.09 0.22 0.23 0.26 0.51\n 0.18]\n4 [ 4 11 115 135 211 214 272 401 537 701 849 854\n 869 1427 1580 1589 1901 1993 2020 2054 2177 2488 2553 2583\n 2643 2674 2714 2969 3286 3641 3866 3875 3923 4016 4042 4379\n 4565 4708 4728 4836 4861 4943 5155 2113 5346 5438 5500 5526\n 5615 5645 5652 5837 5904 5910 6152 6242 6367 6714 6861 6886\n 6887 6967 6977 7086 7236 7399 7418 7448 7453 4791 4799 7841\n 7869 7880 7884 8073 8115 8255 8434 8509 8534 8581 8832 9135\n 6250 9251 9438 9559 6705 9806 9907 9980 10011 10150 10433 10520\n 10553 10667 10754 7999 10915 10966 11101 11110 11277 11303 11400 5622\n 11508 11683 11901 9207 12085 12144 12328 12436 12501 12518 12723 4094\n 1053 7192 12958 12976 13075 13079 13184 4610 13288 13394 13455 13527\n 13626 13637 13665 13718 14316 14329 14389 14395 14496 14565 14625 14691\n 14899 15044 15061 15134 15153 15208 15473 10234 15613 15696 15717 15737\n 15800 15811 15929 16227 16668 14256 17233 17308 17341 12287 17397 17413\n 17489 17667 12769 17890 17901 17972 15673 10383 18184 18229 18347 4851\n 18476 4928 18504 18818 8423 18961 19004 2694 8931 19323 19449 3197\n 148 19903 19932 20081 20169 20283 20325 20563 20650 13477 20791 20806\n 20863 21101 21127 21196 21366 21433 21460 21549 21612 21619 21849 67\n 22025 22059 22066 22085 22122 22138 6947 22427 22445 22607 22838 23093\n 16290 23375 23379 23614 23658 23687 5803 23739 24060 24150 12271 24229\n 24323 913 942 24783 24858 25041 25151 23081 25189 25247 25264 25340\n 25381 25452 25577 8711 25803 25838 14328 25846 25877 3069 26022 12009\n 9219 12066 145 26194 26202 26215 14814 12247 26338 26435 26489 26563\n 26597 26697 17807 26756 26813 26884 1468 18422 27243 18454 27293 27326\n 27409 18843 2486 27746 5698 27841 5757 11591 5938 28139 28293 28309\n 26286 14975 28469 12554 28546 28562 28595 28725 28788 28829 29026 29156\n 29217 27262 10875 29342 29407 29511 29605 29635 18938 29641 29654 29678\n 19058 29787 21486 29854 16939 29943 29955 29965 30246 6717 26496 847\n 30710 30752 30788 30914 5109 31306 11053 23383 31475 31480 31506 8460\n 2542 31653 5736 31697 31758 31816 31837 31893 17038 32028 26204 19643\n 32203 32324 26492 32406 32408 3941 26608 4020 32643 32793 15739 4711\n 10563 32971 32977 20602 33010 1611 33022 1674 33069 33095 33237 1975\n 23286 33365 33372 33511 33540 14140 29736 33600 33671 33708 33734 33745\n 33856 33934 33950 32081 3398 34078 34108 9509 34265 6885 28612 34385\n 34757 34801 1573 34894 1732 34983 10794 35006 31206 1879 35055 35113\n 27415 8146 35199 23409 35299 35415 35430 35464 35507 8881 35670 35693\n 35710 11918 35811 28226 35966 36155 36198 36243 10036 36318 28771 36331\n 36349 4406 10302 32805 36593 32980 36665 36679 34971 36768 33146 23172\n 36859 5148 10999 36892 5233 37025 37028 37046 37141 37186 2685 37337\n 37360 5905 19239 2912 37712 37763 19910 12556 37887 37899 37914 37923\n 37939 17724 978 38140 38236 38270 38351 38379 38385 34838 38491 33246\n 38709 11160 38803 16525 38942 38972 14234 39149 9073 39421 12282 39470\n 692 39643 38006 39730 39819 39884 15761 40024 40042 40128 4892 40251\n 40273 27460 40487 40693 40768 40771 40880 40917 40957 26185 3451 24322\n 41207 41210 41451 41461 12835 41615 41665 41775 41857 41879 41928 41965\n 27413 13747 42096 42211 27667 42492 42515 14447 42735 42792 9414 42853\n 42896 24473 43016 43125 12859 43275 43350 43401 10491 43866 43887 43952\n 44094 16871 44156 44185 40906 44341 39555 605 44619 44655 24577 44921\n 1466 43422 32979 1578 7923 45198 45216 45289 45373 8290 45637 45678\n 19258 45746 6060 45779 24010 45866 21997 17371 46297 46331 10232 46433\n 46473 46494 46495 46506 29102 22967 7777 46700 16404 46903 18946 47022\n 38954 47049 47080 47156 37401 19369 47429 47674 47761 47812 47864 7314\n 1280 48052 29125 31068 7806 48221 48270 13813 48393 11225 8328 48453\n 48471 48528 48578 48584 2658 9057 6440 48950 48966 32321 19919 49224\n 15438 49348 49364 49375 10342 49422 49474 49477 49485 15861 49516 49542\n 49556 31065 49741 49752 49892 49910 49914 18921 49986 50041 50312 100\n 193 209 3781 20160 983 50852 50956 50995 51129 51234 51275 51359\n 51424 35606 51568 11750 28162 51844 51845 51909 34248 51975 52018 52034\n 43088 26660 52110 52117 52156 15568 52262 52327 4740 10580 52415 52437\n 52453 52457 16263 52534 52607 25613 11392 27783 52753 14152 2790 2796\n 42529 28042 52924 50297 19541 12130 19633 53120 3411 374 53196 53208\n 53238 30422 53285 3996 53447 53522 18015 1584 4882 40246 1805 53832\n 43622 49727 18660 8116 2196 54082 54187 5988 54298 33831 24346 12547\n 17615 37948 54899 15688 54930 55076 36713 29373 55238 55247 36905 29553\n 13905 55366 55423 55700 48851 55741 123 51800 55883 55891 562 3728\n 26619 4000 46230 56050 15380 4289 56157 56170 7316 56250 10454 22850\n 22861 56339 56392 56411 7861 56487 56621 38748 56647 56659 56756 56758\n 44295 56996 57040 35819 19557 53138 47486 12504 57297 57315 57389 4124\n 17865 57528 4723 55069 57701 57727 56437 49790 25403 23572 42360 58134\n 58189 28004 21649 58217 48758 58265] [0.03 1. 0.62 0.29 1. 0.12 0.05 1. 0.18 0.19 0.02 0.01 0.05 0.27\n 0.43 0.42 0.03 0.43 0.11 1. 0.33 0.24 0.01 1. 0.36 0.04 1. 0.02\n 0.07 0.41 0.12 0.25 0.05 0.03 0.31 0.59 0.01 0.22 0.13 0.38 1. 0.02\n 0.14 1. 0. 0.01 0.04 0.08 0.23 0.22 0.12 0.2 0.57 0.26 0.05 0.21\n 0.88 0.22 0.71 0.57 0. 0.15 0.13 0.07 0.5 0.28 0.03 0.06 0.32 0.18\n 0.28 0.42 1. 0.13 0.14 0.04 0.14 0.22 1. 0.03 0.22 0.03 0.04 0.29\n 0.02 0. 0.41 0.19 0.49 0.15 0.35 0.11 0.23 0.26 0.55 0.01 0.33 0.05\n 0.23 0.19 0.46 0.61 0.35 0.39 0.33 0.2 0.83 0.17 0.07 0.02 0.15 0.05\n 0.4 0.11 0.61 0.15 0.2 0.54 0.01 0.1 0.13 0.01 0.05 0. 0.52 0.06\n 0.26 0.22 0.09 0.04 0.09 0.01 0. 0.01 0.04 0.35 0.24 0.13 0.28 0.78\n 0.23 0. 1. 0.04 0.22 0.47 0.57 0.19 1. 0.1 0.26 0.03 0.29 1.\n 0.18 0.1 0.35 0.04 0.09 0.08 0.02 0.29 0.18 0.09 0.02 0.41 1. 0.4\n 0.71 0.03 0.06 0.01 0.11 0.05 0.15 0.53 0.14 0.25 0.04 0.05 0.5 0.21\n 0.03 0.06 1. 0.03 0.03 0.16 0.59 0.08 0.34 0.41 0.11 0.12 0.18 0.1\n 0.39 0.05 0.34 0.02 0.19 0.18 0. 0.23 0.33 0.16 1. 0.68 0.06 0.13\n 1. 0.32 0.02 0.26 0.31 0.2 0.36 0.49 0. 0.02 0.1 0.05 1. 0.52\n 0.05 0.26 0.27 0.85 0.16 0.05 0.11 0.05 0.1 0.72 0.05 0.13 0.2 0.04\n 0.13 0.19 0.05 0.08 0.36 0.01 0.86 0.22 0.03 0.55 0.05 0.28 0.25 0.55\n 0.41 0.43 0.31 0.39 0.19 0.11 0.37 0.2 0.56 0.42 1. 0.59 0.27 0.11\n 0.25 0.02 0.2 0.4 0.01 0.33 0.03 0.02 0.12 0.32 0.07 0.08 0.04 0.13\n 0.05 0.43 0.01 1. 0.41 0.74 0.5 0.01 0.14 0.07 0.5 1. 0.7 0.35\n 0.15 0.32 1. 0.07 0.14 1. 0.06 0.62 0.04 0.06 0.47 0.04 0.24 0.03\n 0.26 0.01 1. 0.02 0. 0.36 0.29 0.02 0.16 0.15 0.1 0.04 0.24 0.41\n 0.4 0.44 0.04 0.13 0.21 0.31 0.34 0.25 0.37 0.02 0.03 0.8 0.03 0.18\n 1. 0.24 0.18 0.1 0.4 0.24 0.03 0.35 0.71 1. 0.2 0.36 0.66 0.\n 1. 0.64 0.26 0.24 0.56 0.5 0.18 0.03 0.24 0.65 0.04 0.02 0.26 0.09\n 1. 0.03 0.29 0.12 1. 0.16 0.01 0.62 0.25 0. 0.41 0.4 0.18 0.01\n 0.65 0.22 0.04 0.07 0.31 0.53 0.43 0.26 0.52 0.22 0.01 0.58 0.03 0.23\n 0.42 0.13 0.32 0.06 0.17 0.19 0.04 0.37 0.12 0.58 0.08 0.27 0. 0.35\n 0.34 0.48 0.33 0.1 1. 0.14 0.07 0.3 0.21 0. 0.08 0.24 0.38 0.52\n 0.05 1. 0.09 0.17 0.24 0.44 0.47 0.34 0.33 0.18 0.28 0.01 0.34 0.08\n 0.29 0.37 0.6 0.05 0.5 0.15 0.27 0.08 0.2 0.63 0.45 0.23 0.03 0.1\n 0.01 0.05 0.71 0.2 0.1 0.38 0.43 0.42 0.09 0. 0.1 0.02 0.64 0.5\n 1. 0.09 0.23 0.34 0.04 0.3 0.07 0.1 0.44 1. 0.04 0.02 1. 0.37\n 0.34 0.22 0.14 0.02 0.38 1. 0.07 0.39 0.31 0.41 0.15 0.02 0.87 0.09\n 0.04 0.09 0.15 0.67 0.01 0.31 0.33 0.56 0.53 0.07 0.29 0.85 0.41 0.07\n 0.27 0.01 0.1 0.12 0.03 0.69 0.28 0.04 0.09 0.03 0.39 0.29 0.6 1.\n 0.17 0.01 0.06 0.88 0.17 0.02 0.37 0.47 1. 1. 0.36 0.06 0.4 0.01\n 0.35 0.12 0.44 0.02 0.05 0.56 0.17 0.2 0.04 1. 0.14 0.26 0.59 0.14\n 0.28 0.04 0.25 0.19 0.69 0.43 0.01 0.3 0.09 0.34 0.14 0.21 0.27 0.05\n 0.26 0.43 0.8 0.51 0.42 0.05 0.1 0.03 0.25 0.05 0.19 0.36 0.08 0.27\n 1. 0.01 0.08 0.16 0.17 0.11 0.08 0.11 0.04 0.51 0.09 0.24 0.6 0.02\n 0.03 0.02 0.21 0.03 0.29 0.68 0.19 0.38 0.05 0.03 1. 0.39 0.01 0.\n 1. 0.34 0.77 0.46 0.88 0.06 0.01 0.09 0.32 0.22 0.55 0.16 0.25 0.25\n 0.91 0.6 0.25 0.14 0.25 0.37 0.24 0.12 0.03 0.28 0.32 0.42 0.39 0.16\n 0.27 0.09 0.16 0.19 0.43 0.06 0.08 0.07 0.03 0.39 0.64 0.11 0.28 0.04\n 0.22 0.23 0.05 0.08 0.01 0.16 0.08 0.39 0.26 0.1 0.09 1. 0.05 0.36\n 0.04 0.14 0.35 0.35 0.31 0.54 0. 1. 0.16 0.17 0.11 0.03 0.13 0.23\n 0.2 0.68 0.09 1. 0.08 0.33 1. 0.06 0.03 0.59 0.02 0.35 0.26 0.29\n 0.46 1. 0.71 0.41 0.01 0.22 0.64 0.06 0.23 0.15 0.32 0.38 0.38 0.07\n 0.04 0.97 0.19 0.21 0.94 0.4 0.14 0.05 0.02 0.26 0.29 0.06 0.04 0.02\n 0.36 0.13 0.58 0.18 0.91 0.75 0.01 0.51 0.23 0.01 0.3 0.03 0.34 0.02\n 0.09 0.2 0.25 0.77 0.72 0.24 0.01 0.19 0.62 0.11 0.05 0.36 0.3 0.89\n 0.44 0.23 0.81 0.77 0.46 0.01 0.01 0.39 0.15 0.02 0.07 0.11 0.18 0.3\n 0.6 1. 0.32 0.32 0.14 0.33 0.38 0.39 0.01 0.37 0.45 0.14 0.06 0.15\n 0.03 0.64 0.02 0.41 0.18 0.46 0.48 1. 0.02 0.04 0.05 0.42 0.03 0.57\n 0.03 0.89 0.08 0.6 0.29 0. 0.28 0.55 0. 0.74 1. 0.15 0.12 0.06\n 0.08 1. 0.01 1. 0.06 0.27 0.08 0.1 0.03 0.21 0.06 0.2 0.01 0.04\n 0.28 0.1 1. 0.3 0.13 0.02 0.13 0.77 1. 0.15 0.03 0.7 0.45 0.12\n 0.06 0.32 0.43 0.53 0.06 0.3 0. 0.03]\n5 [ 5 32 35 148 363 442 504 609 712 716 1332 1341\n 1378 1449 2196 2588 3374 3386 430 4312 4409 4562 4605 4678\n 5004 5150 5344 5808 5960 6131 6307 7093 1039 7298 4441 7523\n 7585 7591 7787 7809 7810 8081 8096 8642 8867 8968 9073 9148\n 9252 9364 3641 9612 9744 9908 10014 10046 10168 10360 4568 10552\n 5101 11053 11139 11625 11864 84 12479 12668 13220 13295 13588 13704\n 13840 11230 14441 14520 14992 15136 15159 12687 15293 15407 15495 15499\n 15563 15696 15962 16071 5173 16292 16331 16429 11331 16591 16628 16766\n 5850 11831 17033 17187 17325 17371 397 12373 491 17514 17810 4147\n 17834 17862 17864 17919 18652 18672 18827 18853 18967 19247 14446 19400\n 19513 6608 19884 12660 20025 20143 20158 20359 20567 20757 2042 21115\n 8513 21446 11764 21884 17283 22307 22333 22457 1040 4510 22916 4787\n 2211 23507 23555 23703 21781 24170 24396 6879 10286 25120 25138 16197\n 25333 25336 25445 18806 25485 25499 25916 26238 26331 26387 503 26555\n 20075 4299 26821 26826 4485 22845 27098 27164 13475 27304 13540 27350\n 27568 27714 27884 2749 27910 28063 28399 3595 28593 28644 15673 18378\n 11392 29697 29701 29948 29994 9157 352 22155 30461 7134 30769 10180\n 15560 10216 10256 24823 30874 30904 24985 31491 19097 31743 8868 32274\n 9807 32553 32691 32758 32779 4480 23107 33177 33503 5852 33819 33894\n 34146 15031 34318 34340 9795 32636 15769 32894 35058 35167 35200 8210\n 35295 5486 35355 25614 35471 21661 6118 35745 36015 36020 26531 10030\n 36919 5264 21147 37207 37314 2847 37395 11747 332 7170 7238 15557\n 4570 4623 15936 38428 36713 13507 1843 38615 2070 14139 39040 39349\n 3327 30307 356 34198 39500 39509 38022 39980 39994 40543 8566 41389\n 15848 34880 42376 42506 5994 42529 42567 17077 3254 30141 14945 42874\n 42924 43068 43073 7018 17816 32767 26924 43336 32929 2113 21094 44187\n 3301 44513 44604 37866 9920 44807 28792 1317 45141 45456 12276 15046\n 7013 46329 46352 26844 10675 46685 10934 5164 45537 47046 5741 39072\n 33703 11683 16994 47639 47704 47768 870 47774 47883 47908 47951 48031\n 29253 16191 2279 48494 48615 48617 48763 48907 365 20236 28824 43298\n 49510 49636 29630 3378 50913 18170 15794 51086 38645 21319 19197 51573\n 17109 48818 51733 51736 22144 3969 52020 46235 41543 52237 49460 32903\n 32939 38357 4966 5074 52642 11624 6084 52983 28209 53054 51717 53100\n 53159 53202 26522 39777 15720 39987 53605 53982 54013 2305 5679 5923\n 3528 30606 28809 4669 15967 27425 55390 30030 55740 26330 42950 52107\n 49331 7708 4874 16290 23251 33270 13942 23468 56677 27686 56716 56845\n 28011 56936 56989 17149 19800 7473 54960 57780 51142 57892 57907 57916\n 48467 19009 25706 58104 37435] [0.35 0.02 0.04 1. 0.36 0.2 0.47 0.02 1. 0.12 0.25 0.02 0.31 0.03\n 0.2 0.22 0.14 0.13 0.38 1. 0.31 0.15 0.42 0.44 0.28 0.02 0.58 0.01\n 0.14 0.02 0.02 0.59 0.46 0.48 0.34 0.35 0.28 0.13 0.08 0.09 0.4 0.24\n 0.41 0.04 0.23 0.26 1. 0.32 0.67 0.33 0.24 0.13 0.04 0.24 0.34 0.29\n 0.15 0.63 0.54 0.33 0.59 0.06 0.09 1. 0.44 1. 0.46 0.02 0.14 0.26\n 0.29 0.64 0.19 0.37 0.23 0.12 0.16 0.36 0.2 0.25 0.21 1. 0. 0.25\n 0.12 0.19 0.01 1. 0.34 0.09 0.74 1. 0.02 0.38 0.19 0.21 0.24 0.45\n 0.27 0.17 0.43 0.23 0.21 0.3 0.04 0.16 0.25 0.04 0.22 0.34 0.2 0.07\n 0.35 0.46 0.07 0.26 0.11 0.08 0.11 0.28 1. 0.05 0.2 0.64 0.25 0.44\n 0.04 0.11 0.08 0.26 0.37 1. 0.71 1. 0.33 0.66 0.41 0.38 0.23 0.11\n 0.65 0.24 0.4 0.02 0.53 0.34 1. 1. 0.23 1. 0.54 0.59 0.29 0.07\n 0.59 0.15 0.24 0.3 1. 0.32 0.01 0.14 0.49 0.14 0.17 0.15 0.02 0.75\n 0.5 0.01 0.19 0.14 0.2 0.35 0.31 0.06 0.23 0.21 0.27 0.56 0.05 0.2\n 0.22 0.08 0.34 0.41 0.46 0.46 0.2 0.51 0.41 0.39 0.33 0.29 0.77 0.21\n 0.2 0.25 1. 0.23 0.25 0.01 0.67 1. 0.06 0.38 0.24 0.42 0.21 0.1\n 0.03 0.19 0.03 0.19 1. 0.11 0.89 0.36 0.02 0.17 1. 0.28 0.18 0.29\n 0.12 0. 0.21 0.36 0.08 0.02 0.43 0.03 1. 0.15 0.2 0.22 0.23 0.32\n 0.06 0.18 1. 0.41 1. 0.27 0.03 0.52 1. 0.69 0.02 0.05 0.36 0.15\n 0.06 0.2 0.22 0.03 0.36 0.57 0.44 0.04 0.5 0.37 0.02 0.11 0.08 0.29\n 0.18 0.12 0.01 0. 0.24 0.12 0.14 0.46 0.01 0.26 0.01 0.25 0.29 0.98\n 0.21 0.18 0.08 0.11 0.71 0.23 0.49 0.35 0.24 0.02 0.4 0.14 0.02 0.01\n 0.33 0.26 0.05 0.3 0.01 0.27 0.14 0.12 1. 0.09 0.22 0.23 0.28 0.02\n 0.6 0.01 0.94 0.21 0.19 0.05 0.02 0.6 0.23 0.33 0.16 0.36 0.3 0.\n 0.03 0.23 0.01 1. 0.06 0.05 0.34 0.01 0.5 0.74 0.18 0.22 0.33 0.03\n 0.33 0.14 0.85 0.13 0.29 0.21 0.32 0.35 0.42 0.22 0.1 1. 0.27 0.12\n 0.38 0.4 0.46 0.06 0.25 1. 0.04 0.18 0.32 0.03 0.32 0.16 0.4 0.25\n 0.37 1. 0.02 1. 0.03 0.16 0.11 0.01 0.04 0.16 0.13 0.12 0.31 0.2\n 0.02 0.06 0.02 0.31 0.12 0.25 0.12 0.12 0.39 0.48 0.1 0.01 0.21 0.2\n 0.05 0.79 0.22 0.39 0.96 0. 0.5 0.06 0.12 0.32 0.12 0.77 0.46 0.15\n 0.3 0.94 0.58 0.01 0.04 0.35 0.89 1. 0.16 0.01 1. 0.48 0.68 0.68\n 0.62 0.32 0.13 0.21 0.24 0.45 0.81 1. 0.5 0.29 0.39 1. 1. 0.61\n 1. 0.21 0.29 0.17 0.07 0.17 0.81 0.04 0.03 0.01 0.04 0.39 0.04 0.44\n 0.49]\n" ], [ "import tensorflow as tf\ngrouped_by_items = mapped_df.groupby(\"itemId\")\nwith tf.python_io.TFRecordWriter(\"data/users_for_item\") as ofp:\n for item, grouped in grouped_by_items:\n example = tf.train.Example(features = tf.train.Features(feature = {\n \"key\": tf.train.Feature(int64_list = tf.train.Int64List(value = [item])),\n \"indices\": tf.train.Feature(int64_list = tf.train.Int64List(value = grouped[\"userId\"].values)),\n \"values\": tf.train.Feature(float_list = tf.train.FloatList(value = grouped[\"rating\"].values))\n }))\n ofp.write(example.SerializeToString())", "WARNING:tensorflow:From /tmp/ipykernel_13523/1646577164.py:3: The name tf.python_io.TFRecordWriter is deprecated. Please use tf.io.TFRecordWriter instead.\n\n" ], [ "grouped_by_users = mapped_df.groupby(\"userId\")\nwith tf.python_io.TFRecordWriter(\"data/items_for_user\") as ofp:\n for user, grouped in grouped_by_users:\n example = tf.train.Example(features = tf.train.Features(feature = {\n \"key\": tf.train.Feature(int64_list = tf.train.Int64List(value = [user])),\n \"indices\": tf.train.Feature(int64_list = tf.train.Int64List(value = grouped[\"itemId\"].values)),\n \"values\": tf.train.Feature(float_list = tf.train.FloatList(value = grouped[\"rating\"].values))\n }))\n ofp.write(example.SerializeToString())", "_____no_output_____" ], [ "!ls -lrt data", "total 36580\n-rw-r--r-- 1 jupyter jupyter 14124035 May 16 17:32 collab_raw.csv\n-rw-r--r-- 1 jupyter jupyter 3529114 May 16 17:32 users.csv\n-rw-r--r-- 1 jupyter jupyter 82083 May 16 17:32 items.csv\n-rw-r--r-- 1 jupyter jupyter 7667828 May 16 17:32 collab_mapped.csv\n-rw-r--r-- 1 jupyter jupyter 2296476 May 16 17:33 users_for_item\n-rw-r--r-- 1 jupyter jupyter 9743438 May 16 17:33 items_for_user\n" ] ], [ [ "To summarize, you created the following data files from collab_raw.csv:\n<ol>\n<li> ```collab_mapped.csv``` is essentially the same data as in ```collab_raw.csv``` except that ```visitorId``` and ```contentId``` which are business-specific have been mapped to ```userId``` and ```itemId``` which are enumerated in 0,1,2,.... The mappings themselves are stored in ```items.csv``` and ```users.csv``` so that they can be used during inference.\n<li> ```users_for_item``` contains all the users/ratings for each item in TFExample format\n<li> ```items_for_user``` contains all the items/ratings for each user in TFExample format\n</ol>", "_____no_output_____" ], [ "## Train with WALS\n\nOnce you have the dataset, do matrix factorization with WALS using the [WALSMatrixFactorization](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/factorization/WALSMatrixFactorization) in the contrib directory.\nThis is an estimator model, so it should be relatively familiar.\n<p>\nAs usual, you write an input_fn to provide the data to the model, and then create the Estimator to do train_and_evaluate.\nBecause it is in contrib and hasn't moved over to tf.estimator yet, you use tf.contrib.learn.Experiment to handle the training loop.<p>\n", "_____no_output_____" ] ], [ [ "import os\nimport tensorflow as tf\nfrom tensorflow.python.lib.io import file_io\nfrom tensorflow.contrib.factorization import WALSMatrixFactorization\n \ndef read_dataset(mode, args):\n# TODO 2: Decode the example\n def decode_example(protos, vocab_size):\n features = {\n \"key\": tf.FixedLenFeature(shape = [1], dtype = tf.int64),\n \"indices\": tf.VarLenFeature(dtype = tf.int64),\n \"values\": tf.VarLenFeature(dtype = tf.float32)}\n parsed_features = tf.parse_single_example(serialized = protos, features = features)\n values = tf.sparse_merge(sp_ids = parsed_features[\"indices\"], sp_values = parsed_features[\"values\"], vocab_size = vocab_size)\n # Save key to remap after batching\n # This is a temporary workaround to assign correct row numbers in each batch.\n # You can ignore details of this part and remap_keys().\n key = parsed_features[\"key\"]\n decoded_sparse_tensor = tf.SparseTensor(indices = tf.concat(values = [values.indices, [key]], axis = 0), \n values = tf.concat(values = [values.values, [0.0]], axis = 0), \n dense_shape = values.dense_shape)\n return decoded_sparse_tensor\n \n \n def remap_keys(sparse_tensor):\n # Current indices of your SparseTensor that you need to fix\n bad_indices = sparse_tensor.indices # shape = (current_batch_size * (number_of_items/users[i] + 1), 2)\n # Current values of your SparseTensor that you need to fix\n bad_values = sparse_tensor.values # shape = (current_batch_size * (number_of_items/users[i] + 1),)\n\n # Since batch is ordered, the last value for a batch index is the user\n # Find where the batch index chages to extract the user rows\n # 1 where user, else 0\n user_mask = tf.concat(values = [bad_indices[1:,0] - bad_indices[:-1,0], tf.constant(value = [1], dtype = tf.int64)], axis = 0) # shape = (current_batch_size * (number_of_items/users[i] + 1), 2)\n\n # Mask out the user rows from the values\n good_values = tf.boolean_mask(tensor = bad_values, mask = tf.equal(x = user_mask, y = 0)) # shape = (current_batch_size * number_of_items/users[i],)\n item_indices = tf.boolean_mask(tensor = bad_indices, mask = tf.equal(x = user_mask, y = 0)) # shape = (current_batch_size * number_of_items/users[i],)\n user_indices = tf.boolean_mask(tensor = bad_indices, mask = tf.equal(x = user_mask, y = 1))[:, 1] # shape = (current_batch_size,)\n\n good_user_indices = tf.gather(params = user_indices, indices = item_indices[:,0]) # shape = (current_batch_size * number_of_items/users[i],)\n\n # User and item indices are rank 1, need to make rank 1 to concat\n good_user_indices_expanded = tf.expand_dims(input = good_user_indices, axis = -1) # shape = (current_batch_size * number_of_items/users[i], 1)\n good_item_indices_expanded = tf.expand_dims(input = item_indices[:, 1], axis = -1) # shape = (current_batch_size * number_of_items/users[i], 1)\n good_indices = tf.concat(values = [good_user_indices_expanded, good_item_indices_expanded], axis = 1) # shape = (current_batch_size * number_of_items/users[i], 2)\n\n remapped_sparse_tensor = tf.SparseTensor(indices = good_indices, values = good_values, dense_shape = sparse_tensor.dense_shape)\n return remapped_sparse_tensor\n\n \n def parse_tfrecords(filename, vocab_size):\n if mode == tf.estimator.ModeKeys.TRAIN:\n num_epochs = None # indefinitely\n else:\n num_epochs = 1 # end-of-input after this\n\n files = tf.gfile.Glob(filename = os.path.join(args[\"input_path\"], filename))\n\n # Create dataset from file list\n dataset = tf.data.TFRecordDataset(files)\n dataset = dataset.map(map_func = lambda x: decode_example(x, vocab_size))\n dataset = dataset.repeat(count = num_epochs)\n dataset = dataset.batch(batch_size = args[\"batch_size\"])\n dataset = dataset.map(map_func = lambda x: remap_keys(x))\n return dataset.make_one_shot_iterator().get_next()\n \n def _input_fn():\n features = {\n WALSMatrixFactorization.INPUT_ROWS: parse_tfrecords(\"items_for_user\", args[\"nitems\"]),\n WALSMatrixFactorization.INPUT_COLS: parse_tfrecords(\"users_for_item\", args[\"nusers\"]),\n WALSMatrixFactorization.PROJECT_ROW: tf.constant(True)\n }\n return features, None\n\n return _input_fn", "_____no_output_____" ] ], [ [ "This code is helpful in developing the input function. You don't need it in production.", "_____no_output_____" ] ], [ [ "def try_out():\n with tf.Session() as sess:\n fn = read_dataset(\n mode = tf.estimator.ModeKeys.EVAL, \n args = {\"input_path\": \"data\", \"batch_size\": 4, \"nitems\": NITEMS, \"nusers\": NUSERS})\n feats, _ = fn()\n \n print(feats[\"input_rows\"].eval())\n print(feats[\"input_rows\"].eval())\n\ntry_out()", "WARNING:tensorflow:From /tmp/ipykernel_13523/171886433.py:2: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.\n\nWARNING:tensorflow:From /tmp/ipykernel_13523/3925013467.py:57: The name tf.gfile.Glob is deprecated. Please use tf.io.gfile.glob instead.\n\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/python/autograph/converters/directives.py:119: The name tf.FixedLenFeature is deprecated. Please use tf.io.FixedLenFeature instead.\n\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/python/autograph/converters/directives.py:119: The name tf.VarLenFeature is deprecated. Please use tf.io.VarLenFeature instead.\n\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/python/autograph/converters/directives.py:119: The name tf.parse_single_example is deprecated. Please use tf.io.parse_single_example instead.\n\n" ], [ "def find_top_k(user, item_factors, k):\n all_items = tf.matmul(a = tf.expand_dims(input = user, axis = 0), b = tf.transpose(a = item_factors))\n topk = tf.nn.top_k(input = all_items, k = k)\n return tf.cast(x = topk.indices, dtype = tf.int64)\n \ndef batch_predict(args):\n import numpy as np\n with tf.Session() as sess:\n estimator = tf.contrib.factorization.WALSMatrixFactorization(\n num_rows = args[\"nusers\"], \n num_cols = args[\"nitems\"],\n embedding_dimension = args[\"n_embeds\"],\n model_dir = args[\"output_dir\"])\n \n # This is how you would get the row factors for out-of-vocab user data\n # row_factors = list(estimator.get_projections(input_fn=read_dataset(tf.estimator.ModeKeys.EVAL, args)))\n # user_factors = tf.convert_to_tensor(np.array(row_factors))\n\n # But for in-vocab data, the row factors are already in the checkpoint\n user_factors = tf.convert_to_tensor(value = estimator.get_row_factors()[0]) # (nusers, nembeds)\n # In either case, you have to assume catalog doesn\"t change, so col_factors are read in\n item_factors = tf.convert_to_tensor(value = estimator.get_col_factors()[0])# (nitems, nembeds)\n\n # For each user, find the top K items\n topk = tf.squeeze(input = tf.map_fn(fn = lambda user: find_top_k(user, item_factors, args[\"topk\"]), elems = user_factors, dtype = tf.int64))\n with file_io.FileIO(os.path.join(args[\"output_dir\"], \"batch_pred.txt\"), mode = 'w') as f:\n for best_items_for_user in topk.eval():\n f.write(\",\".join(str(x) for x in best_items_for_user) + '\\n')\n\ndef train_and_evaluate(args):\n train_steps = int(0.5 + (1.0 * args[\"num_epochs\"] * args[\"nusers\"]) / args[\"batch_size\"])\n steps_in_epoch = int(0.5 + args[\"nusers\"] / args[\"batch_size\"])\n print(\"Will train for {} steps, evaluating once every {} steps\".format(train_steps, steps_in_epoch))\n def experiment_fn(output_dir):\n return tf.contrib.learn.Experiment(\n tf.contrib.factorization.WALSMatrixFactorization(\n num_rows = args[\"nusers\"], \n num_cols = args[\"nitems\"],\n embedding_dimension = args[\"n_embeds\"],\n model_dir = args[\"output_dir\"]),\n train_input_fn = read_dataset(tf.estimator.ModeKeys.TRAIN, args),\n eval_input_fn = read_dataset(tf.estimator.ModeKeys.EVAL, args),\n train_steps = train_steps,\n eval_steps = 1,\n min_eval_frequency = steps_in_epoch\n )\n\n from tensorflow.contrib.learn.python.learn import learn_runner\n learn_runner.run(experiment_fn = experiment_fn, output_dir = args[\"output_dir\"])\n \n batch_predict(args)", "_____no_output_____" ], [ "import shutil\nshutil.rmtree(path = \"wals_trained\", ignore_errors=True)\ntrain_and_evaluate({\n \"output_dir\": \"wals_trained\",\n \"input_path\": \"data/\",\n \"num_epochs\": 0.05,\n \"nitems\": NITEMS,\n \"nusers\": NUSERS,\n\n \"batch_size\": 512,\n \"n_embeds\": 10,\n \"topk\": 3\n })", "Will train for 12 steps, evaluating once every 236 steps\nWARNING:tensorflow:From /tmp/ipykernel_13523/1819520735.py:49: run (from tensorflow.contrib.learn.python.learn.learn_runner) is deprecated and will be removed in a future version.\nInstructions for updating:\nUse tf.estimator.train_and_evaluate.\nWARNING:tensorflow:\nThe TensorFlow contrib module will not be included in TensorFlow 2.0.\nFor more information, please see:\n * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md\n * https://github.com/tensorflow/addons\n * https://github.com/tensorflow/io (for I/O related ops)\nIf you depend on functionality not listed there, please file an issue.\n\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/contrib/learn/python/learn/estimators/estimator.py:1180: BaseEstimator.__init__ (from tensorflow.contrib.learn.python.learn.estimators.estimator) is deprecated and will be removed in a future version.\nInstructions for updating:\nPlease replace uses of any Estimator from tf.contrib.learn with an Estimator from tf.estimator.*\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/contrib/learn/python/learn/estimators/estimator.py:427: RunConfig.__init__ (from tensorflow.contrib.learn.python.learn.estimators.run_config) is deprecated and will be removed in a future version.\nInstructions for updating:\nWhen switching to tf.estimator.Estimator, use tf.estimator.RunConfig instead.\nINFO:tensorflow:Using default config.\nINFO:tensorflow:Using config: {'_task_type': None, '_task_id': 0, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fc3e838dfd0>, '_master': '', '_num_ps_replicas': 0, '_num_worker_replicas': 0, '_environment': 'local', '_is_chief': True, '_evaluation_master': '', '_train_distribute': None, '_eval_distribute': None, '_experimental_max_worker_delay_secs': None, '_device_fn': None, '_tf_config': gpu_options {\n per_process_gpu_memory_fraction: 1.0\n}\n, '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_secs': 600, '_log_step_count_steps': 100, '_protocol': None, '_session_config': None, '_save_checkpoints_steps': None, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_model_dir': 'wals_trained', '_session_creation_timeout_secs': 7200}\nWARNING:tensorflow:From /tmp/ipykernel_13523/1819520735.py:45: Experiment.__init__ (from tensorflow.contrib.learn.python.learn.experiment) is deprecated and will be removed in a future version.\nInstructions for updating:\nPlease switch to tf.estimator.train_and_evaluate. You will also have to convert to a tf.estimator.Estimator.\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/contrib/learn/python/learn/monitors.py:279: BaseMonitor.__init__ (from tensorflow.contrib.learn.python.learn.monitors) is deprecated and will be removed after 2016-12-05.\nInstructions for updating:\nMonitors are deprecated. Please use tf.train.SessionRunHook.\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/python/training/training_util.py:236: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.\nInstructions for updating:\nUse Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.\nWARNING:tensorflow:From /opt/conda/lib/python3.7/site-packages/tensorflow_core/contrib/factorization/python/ops/wals.py:315: ModelFnOps.__new__ (from tensorflow.contrib.learn.python.learn.estimators.model_fn) is deprecated and will be removed in a future version.\nInstructions for updating:\nWhen switching to tf.estimator.Estimator, use tf.estimator.EstimatorSpec. You can use the `estimator_spec` method to create an equivalent one.\nINFO:tensorflow:Create CheckpointSaverHook.\nINFO:tensorflow:Graph was finalized.\nINFO:tensorflow:Running local_init_op.\nINFO:tensorflow:Done running local_init_op.\nINFO:tensorflow:Saving checkpoints for 0 into wals_trained/model.ckpt.\nINFO:tensorflow:SweepHook running init op.\nINFO:tensorflow:SweepHook running prep ops for the row sweep.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:loss = 193699.97, step = 1\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Next fit step starting.\nINFO:tensorflow:Saving checkpoints for 12 into wals_trained/model.ckpt.\nINFO:tensorflow:Loss for final step: 162428.39.\nINFO:tensorflow:Starting evaluation at 2022-05-16T17:40:42Z\nINFO:tensorflow:Graph was finalized.\nINFO:tensorflow:Restoring parameters from wals_trained/model.ckpt-12\nINFO:tensorflow:Running local_init_op.\nINFO:tensorflow:Done running local_init_op.\nINFO:tensorflow:Evaluation [1/1]\nINFO:tensorflow:Finished evaluation at 2022-05-16-17:40:43\nINFO:tensorflow:Saving dict for global step 12: global_step = 12, loss = 193699.97\nINFO:tensorflow:Using default config.\nINFO:tensorflow:Using config: {'_task_type': None, '_task_id': 0, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fc3e838dfd0>, '_master': '', '_num_ps_replicas': 0, '_num_worker_replicas': 0, '_environment': 'local', '_is_chief': True, '_evaluation_master': '', '_train_distribute': None, '_eval_distribute': None, '_experimental_max_worker_delay_secs': None, '_device_fn': None, '_tf_config': gpu_options {\n per_process_gpu_memory_fraction: 1.0\n}\n, '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_secs': 600, '_log_step_count_steps': 100, '_protocol': None, '_session_config': None, '_save_checkpoints_steps': None, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_model_dir': 'wals_trained', '_session_creation_timeout_secs': 7200}\n" ], [ "!ls wals_trained", "batch_pred.txt\ncheckpoint\neval\nevents.out.tfevents.1652722839.tensorflow-1-15-20220516-225105\ngraph.pbtxt\nmodel.ckpt-0.data-00000-of-00001\nmodel.ckpt-0.index\nmodel.ckpt-0.meta\nmodel.ckpt-12.data-00000-of-00001\nmodel.ckpt-12.index\nmodel.ckpt-12.meta\n" ], [ "!head wals_trained/batch_pred.txt", "5526,129,885\n4712,2402,2620\n2710,139,1001\n885,5526,3450\n202,370,2839\n4972,1085,4505\n4896,1687,2833\n3418,5095,2702\n3537,479,4896\n1001,4516,2347\n" ] ], [ [ "## Run as a Python module\n\nLet's run it as Python module for just a few steps.", "_____no_output_____" ] ], [ [ "os.environ[\"NITEMS\"] = str(NITEMS)\nos.environ[\"NUSERS\"] = str(NUSERS)", "_____no_output_____" ], [ "%%bash\nrm -rf wals.tar.gz wals_trained\ngcloud ai-platform local train \\\n --module-name=walsmodel.task \\\n --package-path=${PWD}/walsmodel \\\n -- \\\n --output_dir=${PWD}/wals_trained \\\n --input_path=${PWD}/data \\\n --num_epochs=0.01 --nitems=${NITEMS} --nusers=${NUSERS} \\\n --job-dir=./tmp", "Will train for 2 steps, evaluating once every 236 steps\n" ] ], [ [ "## Run on Cloud", "_____no_output_____" ] ], [ [ "%%bash\ngsutil -m cp data/* gs://${BUCKET}/wals/data", "Copying file://data/collab_mapped.csv [Content-Type=text/csv]...\nCopying file://data/collab_raw.csv [Content-Type=text/csv]... \nCopying file://data/items.csv [Content-Type=text/csv]... \nCopying file://data/items_for_user [Content-Type=application/octet-stream]... \nCopying file://data/users_for_item [Content-Type=application/octet-stream]... \nCopying file://data/users.csv [Content-Type=text/csv]... \n- [6/6 files][ 35.7 MiB/ 35.7 MiB] 100% Done \nOperation completed over 6 objects/35.7 MiB. \n" ], [ "%%bash\nOUTDIR=gs://${BUCKET}/wals/model_trained\nJOBNAME=wals_$(date -u +%y%m%d_%H%M%S)\necho $OUTDIR $REGION $JOBNAME\ngsutil -m rm -rf $OUTDIR\ngcloud ai-platform jobs submit training $JOBNAME \\\n --region=$REGION \\\n --module-name=walsmodel.task \\\n --package-path=${PWD}/walsmodel \\\n --job-dir=$OUTDIR \\\n --staging-bucket=gs://$BUCKET \\\n --scale-tier=BASIC_GPU \\\n --runtime-version=$TFVERSION \\\n -- \\\n --output_dir=$OUTDIR \\\n --input_path=gs://${BUCKET}/wals/data \\\n --num_epochs=10 --nitems=${NITEMS} --nusers=${NUSERS} ", "gs://qwiklabs-gcp-01-7f6e984e70dc/wals/model_trained us-central1 wals_220516_174236\njobId: wals_220516_174236\nstate: QUEUED\n" ] ], [ [ "This will take <b>10 minutes</b> to complete. Rerun the above command until the jobs gets submitted.", "_____no_output_____" ], [ "## Get row and column factors\n\nOnce you have a trained WALS model, you can get row and column factors (user and item embeddings) from the checkpoint file. You'll look at how to use these in the section on building a recommendation system using deep neural networks.", "_____no_output_____" ] ], [ [ "def get_factors(args):\n with tf.Session() as sess:\n estimator = tf.contrib.factorization.WALSMatrixFactorization(\n num_rows = args[\"nusers\"], \n num_cols = args[\"nitems\"],\n embedding_dimension = args[\"n_embeds\"],\n model_dir = args[\"output_dir\"])\n \n row_factors = estimator.get_row_factors()[0]\n col_factors = estimator.get_col_factors()[0]\n return row_factors, col_factors", "_____no_output_____" ], [ "args = {\n \"output_dir\": \"gs://{}/wals/model_trained\".format(BUCKET),\n \"nitems\": NITEMS,\n \"nusers\": NUSERS,\n \"n_embeds\": 10\n }\n\nuser_embeddings, item_embeddings = get_factors(args)\nprint(user_embeddings[:3])\nprint(item_embeddings[:3])", "INFO:tensorflow:Using default config.\nINFO:tensorflow:Using config: {'_task_type': None, '_task_id': 0, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fc3e9a79a10>, '_master': '', '_num_ps_replicas': 0, '_num_worker_replicas': 0, '_environment': 'local', '_is_chief': True, '_evaluation_master': '', '_train_distribute': None, '_eval_distribute': None, '_experimental_max_worker_delay_secs': None, '_device_fn': None, '_tf_config': gpu_options {\n per_process_gpu_memory_fraction: 1.0\n}\n, '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_secs': 600, '_log_step_count_steps': 100, '_protocol': None, '_session_config': None, '_save_checkpoints_steps': None, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_model_dir': 'gs://qwiklabs-gcp-01-7f6e984e70dc/wals/model_trained', '_session_creation_timeout_secs': 7200}\n[[ 3.6948849e-03 -8.5906551e-04 4.8303331e-04 -1.5890118e-03\n 3.2951573e-03 1.8742392e-03 -4.0969993e-03 6.1582108e-03\n -2.8773610e-04 -1.3332820e-03]\n [ 1.3797252e-03 -6.1344338e-04 1.8084486e-03 -1.3052947e-03\n -5.8687606e-04 1.2303573e-03 4.5222897e-04 1.5270878e-03\n -3.6727260e-03 -3.0614049e-04]\n [-1.8781899e-04 -3.4702761e-04 -2.0322317e-05 -3.4224652e-04\n -4.4325300e-04 -1.1415767e-03 1.1274662e-04 5.7519245e-04\n 1.1694311e-04 7.0931134e-04]]\n[[ 1.55777712e+01 -1.05091429e+00 9.74276304e-01 -1.42321882e+01\n 5.99148512e+00 2.02531338e+00 -1.14337492e+01 1.73737488e+01\n -6.27120638e+00 -3.14125824e+00]\n [ 1.13881841e-01 -1.24881916e-01 1.47017851e-01 3.90907042e-02\n -2.54981425e-02 -2.25878507e-02 -8.38336051e-02 -3.10666151e-02\n -1.92982569e-01 5.66481380e-03]\n [-2.36634836e-01 -1.71446967e+00 -2.04065472e-01 -2.04369950e+00\n -1.53748310e+00 -4.64286995e+00 1.42189160e-01 2.47772884e+00\n -9.40964222e-01 3.68809652e+00]]\n" ] ], [ [ "You can visualize the embedding vectors using dimensional reduction techniques such as PCA.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nfrom mpl_toolkits.mplot3d import Axes3D\nfrom sklearn.decomposition import PCA\n\npca = PCA(n_components = 3)\npca.fit(user_embeddings)\n# TODO 3: Apply the mapping (transform) to user embeddings\nuser_embeddings_pca = pca.transform(user_embeddings)\n\nfig = plt.figure(figsize = (8,8))\nax = fig.add_subplot(111, projection = \"3d\")\nxs, ys, zs = user_embeddings_pca[::150].T\nax.scatter(xs, ys, zs)", "_____no_output_____" ] ], [ [ "<pre>\n# Copyright 2022 Google Inc. All Rights Reserved.\n#\n# Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# http://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.\n</pre>", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "import numpy as np\nfrom resonance.nonlinear_systems import SingleDoFNonLinearSystem", "_____no_output_____" ] ], [ [ "To apply arbitrary forcing to a single degree of freedom linear or nonlinear system, you can do so with `SingleDoFNonLinearSystem` (`SingleDoFLinearSystem` does not support arbitrary forcing...yet).\n\nAdd constants, a generalized coordinate, and a generalized speed to the system.", "_____no_output_____" ] ], [ [ "sys = SingleDoFNonLinearSystem()", "_____no_output_____" ], [ "sys.constants['m'] = 100 # kg\nsys.constants['c'] = 1.1*1.2*0.5/2\nsys.constants['k'] = 10\nsys.constants['Fo'] = 1000 # N\nsys.constants['Ft'] = 100 # N/s\nsys.constants['to'] = 3.0 # s\n\nsys.coordinates['x'] = 0.0\nsys.speeds['v'] = 0.0", "_____no_output_____" ] ], [ [ "Create a function that evaluates the first order form of the non-linear equations of motion. In this case:\n\n$$\n\\dot{x} = v \\\\\nm\\dot{v} + c \\textrm{sgn}(v)v^2 + k \\textrm{sgn}(x)x^2 = F(t)\n$$\n\nMake the arbitrary forcing term, $F$, an input to this function.", "_____no_output_____" ] ], [ [ "def eval_eom(x, v, m, c, k, F):\n xdot = v\n vdot = (F - np.sign(v)*c*v**2 - np.sign(x)*k*x**2) / m\n return xdot, vdot", "_____no_output_____" ] ], [ [ "Note that you cannot add this to the system because `F` has not been defined.", "_____no_output_____" ] ], [ [ "sys.diff_eq_func = eval_eom", "_____no_output_____" ] ], [ [ "To rememdy this, create a function that returns the input value given the appropriate constants and time.", "_____no_output_____" ] ], [ [ "def eval_step_input(Fo, to, time):\n if time < to:\n return 0.0\n else:\n return Fo", "_____no_output_____" ], [ "import matplotlib.pyplot as plt\n%matplotlib widget", "_____no_output_____" ], [ "ts = np.linspace(0, 10)\nplt.plot(ts, eval_step_input(5.0, 3.0, ts))", "_____no_output_____" ], [ "ts < 3.0", "_____no_output_____" ], [ "def eval_step_input(Fo, to, time):\n F = np.empty_like(time)\n for i, ti in enumerate(time):\n if ti < to:\n F[i] = 0.0\n else:\n F[i] = Fo\n return F", "_____no_output_____" ], [ "plt.plot(ts, eval_step_input(5.0, 3.0, ts))", "_____no_output_____" ], [ "eval_step_input(5.0, 3.0, ts)", "_____no_output_____" ], [ "eval_step_input(5.0, 3.0, 7.0)", "_____no_output_____" ], [ "def eval_step_input(Fo, to, time):\n if np.isscalar(time):\n if time < to:\n return 0.0\n else:\n return Fo\n else:\n F = np.empty_like(time)\n for i, ti in enumerate(time):\n if ti < to:\n F[i] = 0.0\n else:\n F[i] = Fo\n return F", "_____no_output_____" ], [ "eval_step_input(5.0, 3.0, 7.0)", "_____no_output_____" ], [ "eval_step_input(5.0, 3.0, ts)", "_____no_output_____" ], [ "True * 5.0", "_____no_output_____" ], [ "False * 5.0", "_____no_output_____" ], [ "(ts >= 3.0)*5.0", "_____no_output_____" ], [ "(5.0 >= 3.0)*5.0", "_____no_output_____" ], [ "def eval_step_input(Fo, to, time):\n return (time >=to)*Fo", "_____no_output_____" ], [ "eval_step_input(5.0, 3.0, ts)", "_____no_output_____" ], [ "eval_step_input(5.0, 3.0, 7.0)", "_____no_output_____" ], [ "sys.add_measurement('F', eval_step_input)", "_____no_output_____" ], [ "sys.diff_eq_func = eval_eom", "_____no_output_____" ], [ "traj = sys.free_response(20.0)", "_____no_output_____" ], [ "traj.plot(subplots=True)", "_____no_output_____" ], [ "def eval_ramp_input(Ft, to, time):\n return (time >= to)*(Ft*time - Ft*to)", "_____no_output_____" ], [ "del sys.measurements['F']", "_____no_output_____" ], [ "sys.add_measurement('F', eval_ramp_input)", "_____no_output_____" ], [ "sys.measurements", "_____no_output_____" ], [ "traj = sys.free_response(20.0)", "_____no_output_____" ], [ "traj.plot(subplots=True)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Fit $k_{ij}$ and $r_c^{ABij}$ interactions parameter of Ethanol and CPME\n\nThis notebook has te purpose of showing how to optimize the $k_{ij}$ and $r_c^{ABij}$ for a mixture with induced association.\n\nFirst it's needed to import the necessary modules", "_____no_output_____" ] ], [ [ "import numpy as np\n\nfrom sgtpy import component, mixture, saftvrmie\nfrom sgtpy.fit import fit_cross", "_____no_output_____" ] ], [ [ "Now that the functions are available it is necessary to create the mixture.", "_____no_output_____" ] ], [ [ "ethanol = component('ethanol2C', ms = 1.7728, sigma = 3.5592 , eps = 224.50,\n lambda_r = 11.319, lambda_a = 6., eAB = 3018.05, rcAB = 0.3547,\n rdAB = 0.4, sites = [1,0,1], cii= 5.3141080872882285e-20)\n\ncpme = component('cpme', ms = 2.32521144, sigma = 4.13606074, eps = 343.91193798, lambda_r = 14.15484877, \n lambda_a = 6.0, npol = 1.91990385,mupol = 1.27, sites =[0,0,1], cii = 3.5213681817448466e-19)\n\nmix = mixture(ethanol, cpme)", "_____no_output_____" ] ], [ [ "Now the experimental equilibria data is read and a tuple is created. It includes the experimental liquid composition, vapor composition, equilibrium temperature and pressure. This is done with ```datavle = (Xexp, Yexp, Texp, Pexp)```\n", "_____no_output_____" ] ], [ [ "# Experimental data obtained from Mejia, Cartes, J. Chem. Eng. Data, vol. 64, no. 5, pp. 1970–1977, 2019\n\n# Experimental temperature saturation in K\nTexp = np.array([355.77, 346.42, 342.82, 340.41, 338.95, 337.78, 336.95, 336.29,\n 335.72, 335.3 , 334.92, 334.61, 334.35, 334.09, 333.92, 333.79,\n 333.72, 333.72, 333.81, 334.06, 334.58])\n\n# Experimental pressure in Pa\nPexp = np.array([50000., 50000., 50000., 50000., 50000., 50000., 50000., 50000.,\n 50000., 50000., 50000., 50000., 50000., 50000., 50000., 50000.,\n 50000., 50000., 50000., 50000., 50000.])\n\n# Experimental liquid composition\nXexp = np.array([[0. , 0.065, 0.11 , 0.161, 0.203, 0.253, 0.301, 0.351, 0.402,\n 0.446, 0.497, 0.541, 0.588, 0.643, 0.689, 0.743, 0.785, 0.837,\n 0.893, 0.947, 1. ],\n [1. , 0.935, 0.89 , 0.839, 0.797, 0.747, 0.699, 0.649, 0.598,\n 0.554, 0.503, 0.459, 0.412, 0.357, 0.311, 0.257, 0.215, 0.163,\n 0.107, 0.053, 0. ]])\n\n# Experimental vapor composition\nYexp = np.array([[0. , 0.302, 0.411, 0.48 , 0.527, 0.567, 0.592, 0.614, 0.642,\n 0.657, 0.678, 0.694, 0.71 , 0.737, 0.753, 0.781, 0.801, 0.837,\n 0.883, 0.929, 1. ],\n [1. , 0.698, 0.589, 0.52 , 0.473, 0.433, 0.408, 0.386, 0.358,\n 0.343, 0.322, 0.306, 0.29 , 0.263, 0.247, 0.219, 0.199, 0.163,\n 0.117, 0.071, 0. ]])\n\ndatavle = (Xexp, Yexp, Texp, Pexp)", "_____no_output_____" ] ], [ [ "The function ```fit_cross``` optimize the $k_{ij}$ correction and $r_c^{ABij}$ distance. An initial guess is needed, as well as the mixture object, the index of the self-associating component and the equilibria data.", "_____no_output_____" ] ], [ [ "#initial guesses for kij and rcij\nx0 = [0.01015194, 2.23153033]\nfit_cross(x0, mix, assoc=0, datavle=datavle)", "_____no_output_____" ] ], [ [ "For more information just run:\n```fit_cross?```", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# BAYES CLASSIFIERS\n\nFor any classifier $f:{X \\to Y}$, it's prediction error is:\n\n$P(f(x) \\ne Y) = \\mathbb{E}[ \\mathbb{1}(f(X) \\ne Y)] = \\mathbb{E}[\\mathbb{E}[ \\mathbb{1}(f(X) \\ne Y)|X]]$\n\nFor each $x \\in X$,\n\n$$\\mathbb{E}[ \\mathbb{1}(f(X) \\ne Y)|X = x] = \\sum\\limits_{y \\in Y} P(Y = y|X = x) \\cdot \\mathbb{1}(f(x) \\ne y)$$\n\nThe above quantity is minimized for this particular $x \\in X$ when,\n\n$$f(x) = \\underset{y \\in Y}{argmax} \\space P(Y = y|X = x) \\space \\star$$\n\nA classifier $f$ with property $ \\star$ for all $x \\in X$ is called the `Bayes Classifier`\n", "_____no_output_____" ], [ "Under the assumption $(X,Y) \\overset{iid}{\\sim} P$, the optimal classifier is:\n$$f^{\\star}(x) = \\underset{y \\in Y}{argmax} \\space P(Y = y|X = x)$$\n\nAnd from _Bayes Rule_ we equivalently have:\n\n$$f^{\\star}(x) = \\underset{y \\in Y}{argmax} \\space P(Y = y) \\space P(X = x|Y = y)$$\n\nWhere\n- $P(Y =y)$ is called _the class prior_\n- $P(X = x|Y= y)$ is called _the class conditional distribution_ of $X$\n\nAssuming $X = \\mathbb{R}, Y = \\{ 0,1 \\}$, and the distribution of $P \\space \\text{of} \\space (X,Y)$ is as follows:\n- _Class prior_: $P(Y = y) = \\pi_y, y \\in \\{ 0,1 \\}$\n- _Class conditional density_ for class $y \\in \\{ 0,1 \\}: p_y (x) = N(x|\\mu_y,\\sigma^2_y)$\n\n$$f^{\\star}(x) = \\underset{y \\in \\{ 0,1 \\}}{argmax} \\space P(Y = y) \\space P(X = x|Y = y) = \n \\begin{cases}\n 1 & \\text{if} \\space \\frac{\\pi_1}{\\sigma_1}\\space exp[- \\frac{(x - \\mu_1)^2}{2 \\sigma^2_1}] > \\frac{\\pi_0}{\\sigma_0}\\space exp[- \\frac{(x - \\mu_0)^2}{2 \\sigma^2_0}]\\\\\n 0 & \\text{otherwise}\n \\end{cases}$$", "_____no_output_____" ], [ "### _Bayes Classifier_\n", "_____no_output_____" ], [ "The `Bayes Classifier` has the smallest prediction error of all classifiers. The problem is that we need to know the distribution of $P$ in order to construct the `Bayes Classifier`", "_____no_output_____" ], [ "# NAIVE BAYES CLASSIFIER\n\nA simplifying assumtion that the features values are conditionally independent given the label, the probability of observing the conjunction $x_1, x_2, x_3, ..., x_d$ is the product of the probabilities for the individual features:\n\n$$ p(x_1, x_2, x_3, ..., x_d|y) = \\prod \\limits_j \\space p(x_j|y)$$\n\nThen the `Naive Bayes Classifier` is defined as:\n\n$$f^{\\star}(x) = \\underset{y \\in Y}{argmax} \\space p(y) \\space \\prod \\limits_j \\space p(x_j|y)$$\n\nWe can estimate these two terms based on the **frequency counts** in the dataset. If the features are real-valued, Naive Bayes can be extended assuming that features follow a Gaussian distribution. This extension is called `Gaussian Naive Bayes`. Other functions can be used to estimate the distribution but the Gaussian distribution is the easiest to work with due to we only need to estimate the mean and the standard deviation from the dataset.\n\nOk, let's start with the implementation of `Gaussian Naive Bayes` from scratch.", "_____no_output_____" ] ], [ [ "##IMPORTING ALL NECESSARY SUPPORT LIBRARIES\nimport math as mt\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n%matplotlib inline", "_____no_output_____" ], [ "def separate_by_label(dataset):\n separate = dict()\n for i in range(len(dataset)):\n row = dataset[i]\n label = row[-1]\n if (label not in separate):\n separate[label] = list()\n separate[label].append(row)\n \n return separate", "_____no_output_____" ], [ "def mean(list_num):\n return sum(list_num)/len(list_num)", "_____no_output_____" ], [ "def stdv(list_num):\n mu = mean(list_num)\n var = sum([(x - mu)**2 for x in list_num])/(len(list_num) - 1)\n \n return mt.sqrt(var)", "_____no_output_____" ], [ "def stats_per_feature(ds):\n '''\n argument:\n > ds: 1-D Array with the all data separated by class\n returns:\n > stats: 1-D Array with statistics summary for each feature\n '''\n stats = [(mean(col), stdv(col), len(col)) for col in zip(*ds)]\n del(stats[-1])\n \n return stats", "_____no_output_____" ], [ "def summary_by_class(dataset):\n sep_label = separate_by_label(dataset)\n summary = dict()\n for label, rows in sep_label.items():\n summary[label] = stats_per_feature(rows)\n \n return summary", "_____no_output_____" ], [ "def gaussian_pdf(mean, stdv, x):\n _exp = mt.exp(-1*((x - mean)**2/(2*stdv**2)))\n return (1/(mt.sqrt(2 * mt.pi)*stdv)) * _exp", "_____no_output_____" ] ], [ [ "Now it is time to use the statistics calculated from the data to calculate probabilities for new data.\n\nProbabilities are calculated separately for each class, so we calculate the probability that a new piece of data belongs to the first class, then calculate the probability that it belongs to the second class, and so on for all the classes.\n\nFor example, if we have two inputs $x_1 and \\space x_2$ the calculation of the probability that those belong to class = _y_ is:\n\n$$P(class = y|x_1,x_2) = P(x_1|class = y) \\cdot P(x_2|class = y) \\cdot P(class = y)$$", "_____no_output_____" ] ], [ [ "def class_probabilities(summary, row):\n total = sum([summary[label][0][2] for label in summary])\n probabilities = dict()\n \n for class_, class_summary in summary.items():\n probabilities[class_] = summary[class_][0][2]/total\n for i in range(len(class_summary)):\n mean, stdev, count = class_summary[i]\n probabilities[class_] *= gaussian_pdf(row[i], mean, stdev)\n \n return probabilities", "_____no_output_____" ], [ "def predict(summary, row):\n cls_prob = class_probabilities(summary, row)\n _label, _prob = None, -1.0\n for class_, probability in cls_prob.items():\n if _label is None or probability > _prob:\n _prob = probability\n _label = class_\n \n return _label ", "_____no_output_____" ] ], [ [ "In order to verify proper implementation a **toy dataset** is used to evaluate the algorithm.", "_____no_output_____" ] ], [ [ "dataset = [[3.393533211,2.331273381,0],\n[3.110073483,1.781539638,0],\n[1.343808831,3.368360954,0],\n[3.582294042,4.67917911,0],\n[2.280362439,2.866990263,0],\n[7.423436942,4.696522875,1],\n[5.745051997,3.533989803,1],\n[9.172168622,2.511101045,1],\n[7.792783481,3.424088941,1],\n[7.939820817,0.791637231,1]]\n\nsummaries = summary_by_class(dataset)\nfor row in dataset:\n y_pred = predict(summaries, row)\n y_real = row[-1]\n print(\"Expected={0}, Predicted={1}\".format(y_real, y_pred))", "Expected=0, Predicted=0\nExpected=0, Predicted=0\nExpected=0, Predicted=0\nExpected=0, Predicted=0\nExpected=0, Predicted=0\nExpected=1, Predicted=1\nExpected=1, Predicted=1\nExpected=1, Predicted=1\nExpected=1, Predicted=1\nExpected=1, Predicted=1\n" ] ], [ [ "# _GAUSSIAN NAIVE BAYES APPLICATION_\n\nFrom the `UCI Machine Learning Repository` which contains Iris dataset, we will train our `Gaussian Naive Bayes` model. The Iris dataset is perhaps the best known database to be found in the pattern recognition literature. Fisher's paper is a classic in the field and is referenced frequently to this day.\n\nThe dataset contains 3 classes of 50 instances each, where each class refers to a type of iris plant. One class is linearly separable from the other 2; the latter are not linearly separable from each other.\n\nThe dataset have 150 instances and the following attributes:\n\n 1. sepal length in cm\n 2. sepal width in cm\n 3. petal length in cm\n 4. petal width in cm\n 5. class: \n -- Iris Setosa\n -- Iris Versicolour\n -- Iris Virginica", "_____no_output_____" ], [ "To compare the performance of our _Classifier_ on the **Iris** dataset, a Gaussian Naive Bayes model from `sklearn` will be fit on the dataset and classification report for both models is generated.", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.metrics import classification_report", "_____no_output_____" ], [ "##LOADING 'IRIS' DATASET\ncolumns = ['sepal-len','sepal-wid','petal-len','petal-wid','class']\ndf = pd.read_csv('./data/Iris.csv', names = columns)\ndf.head()", "_____no_output_____" ], [ "df.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 150 entries, 0 to 149\nData columns (total 5 columns):\nsepal-len 150 non-null float64\nsepal-wid 150 non-null float64\npetal-len 150 non-null float64\npetal-wid 150 non-null float64\nclass 150 non-null object\ndtypes: float64(4), object(1)\nmemory usage: 5.9+ KB\n" ] ], [ [ "Due to the class variable type is `categorical` we need first to encode it as numeric type in order to be feed it into our models.", "_____no_output_____" ] ], [ [ "def encoder(df, class_value_pair):\n for class_name, value in class_value_pair.items():\n df['class'] = df['class'].replace(class_name, value)\n \n return df\nclass_encoder = {'Iris-setosa':0, 'Iris-versicolor':1, 'Iris-virginica':2}\ndf = encoder(df, class_encoder)\ndf.head()", "_____no_output_____" ], [ "df['class'].value_counts().sort_index()", "_____no_output_____" ] ], [ [ "Once the preprocessing is complete the dataset will be split into a `Training` & `Test` dataset.", "_____no_output_____" ] ], [ [ "X_ = df.drop(['class'],axis = 1)\ny = df['class']\nX_train, X_test, y_train, y_test = train_test_split(X_, y, test_size = 0.30, random_state = 5)", "_____no_output_____" ] ], [ [ "Now, we can `train` our customized model. Noticed that our _Gaussian Naive Bayes_ model expects a complete dataset (attributes and labels) in order to calculate the summaries.", "_____no_output_____" ] ], [ [ "ds_train = pd.concat([X_train, y_train], axis = 1)\nGNB_custom = summary_by_class(ds_train.values.tolist())", "_____no_output_____" ], [ "ds_test = pd.concat([X_test, y_test], axis = 1)\ncust_pred = [predict(GNB_custom, row) for row in ds_test.values.tolist()]\ncust_pred = np.array(cust_pred, dtype = 'int64')", "_____no_output_____" ], [ "cust_pred", "_____no_output_____" ] ], [ [ "Now an instance of `sklearn` _Gaussian Naive Bayes_ model is created and fit it with the training data and an array of predictions is obtained in order to get out performance comparation", "_____no_output_____" ] ], [ [ "##GET AND INSTANCE OF GAUSSIAN NAIVE BAYES MODEL\nGNB_skln = GaussianNB()\nGNB_skln.fit(X_train, y_train)\n\n##CREATE SKLEARN PREDICTIONS ARRAY\nsk_pred = GNB_skln.predict(X_test)", "_____no_output_____" ], [ "sk_pred", "_____no_output_____" ] ], [ [ "By last, a comparison on both models is performed thru a _Classification Report_", "_____no_output_____" ] ], [ [ "print(\"Sklearn:\")\nprint(classification_report(y_test, sk_pred))\nprint(\"Custom:\")\nprint(classification_report(y_test, cust_pred))", "Sklearn:\n precision recall f1-score support\n\n 0 1.00 1.00 1.00 15\n 1 0.88 0.94 0.91 16\n 2 0.92 0.86 0.89 14\n\n micro avg 0.93 0.93 0.93 45\n macro avg 0.94 0.93 0.93 45\nweighted avg 0.93 0.93 0.93 45\n\nCustom:\n precision recall f1-score support\n\n 0 1.00 1.00 1.00 15\n 1 0.88 0.94 0.91 16\n 2 0.92 0.86 0.89 14\n\n micro avg 0.93 0.93 0.93 45\n macro avg 0.94 0.93 0.93 45\nweighted avg 0.93 0.93 0.93 45\n\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Mini Project: Temporal-Difference Methods\n\nIn this notebook, you will write your own implementations of many Temporal-Difference (TD) methods.\n\nWhile we have provided some starter code, you are welcome to erase these hints and write your code from scratch.", "_____no_output_____" ], [ "### Part 0: Explore CliffWalkingEnv\n\nUse the code cell below to create an instance of the [CliffWalking](https://github.com/openai/gym/blob/master/gym/envs/toy_text/cliffwalking.py) environment.", "_____no_output_____" ] ], [ [ "import gym\nenv = gym.make('CliffWalking-v0')", "_____no_output_____" ] ], [ [ "The agent moves through a $4\\times 12$ gridworld, with states numbered as follows:\n```\n[[ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11],\n [12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23],\n [24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35],\n [36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47]]\n```\nAt the start of any episode, state `36` is the initial state. State `47` is the only terminal state, and the cliff corresponds to states `37` through `46`.\n\nThe agent has 4 potential actions:\n```\nUP = 0\nRIGHT = 1\nDOWN = 2\nLEFT = 3\n```\n\nThus, $\\mathcal{S}^+=\\{0, 1, \\ldots, 47\\}$, and $\\mathcal{A} =\\{0, 1, 2, 3\\}$. Verify this by running the code cell below.", "_____no_output_____" ] ], [ [ "print(env.action_space)\nprint(env.observation_space)", "Discrete(4)\nDiscrete(48)\n" ] ], [ [ "In this mini-project, we will build towards finding the optimal policy for the CliffWalking environment. The optimal state-value function is visualized below. Please take the time now to make sure that you understand _why_ this is the optimal state-value function.", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom plot_utils import plot_values\n\n# define the optimal state-value function\nV_opt = np.zeros((4,12))\nV_opt[0:13][0] = -np.arange(3, 15)[::-1]\nV_opt[0:13][1] = -np.arange(3, 15)[::-1] + 1\nV_opt[0:13][2] = -np.arange(3, 15)[::-1] + 2\nV_opt[3][0] = -13\n\nplot_values(V_opt)", "_____no_output_____" ] ], [ [ "### Part 1: TD Prediction: State Values\n\nIn this section, you will write your own implementation of TD prediction (for estimating the state-value function).\n\nWe will begin by investigating a policy where the agent moves:\n- `RIGHT` in states `0` through `10`, inclusive, \n- `DOWN` in states `11`, `23`, and `35`, and\n- `UP` in states `12` through `22`, inclusive, states `24` through `34`, inclusive, and state `36`.\n\nThe policy is specified and printed below. Note that states where the agent does not choose an action have been marked with `-1`.", "_____no_output_____" ] ], [ [ "policy = np.hstack([1*np.ones(11), 2, 0, np.zeros(10), 2, 0, np.zeros(10), 2, 0, -1*np.ones(11)])\nprint(\"\\nPolicy (UP = 0, RIGHT = 1, DOWN = 2, LEFT = 3, N/A = -1):\")\nprint(policy.reshape(4,12))", "\nPolicy (UP = 0, RIGHT = 1, DOWN = 2, LEFT = 3, N/A = -1):\n[[ 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 2.]\n [ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 2.]\n [ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 2.]\n [ 0. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1.]]\n" ] ], [ [ "Run the next cell to visualize the state-value function that corresponds to this policy. Make sure that you take the time to understand why this is the corresponding value function!", "_____no_output_____" ] ], [ [ "V_true = np.zeros((4,12))\nfor i in range(3):\n V_true[0:12][i] = -np.arange(3, 15)[::-1] - i\nV_true[1][11] = -2\nV_true[2][11] = -1\nV_true[3][0] = -17\n\nplot_values(V_true)", "_____no_output_____" ] ], [ [ "The above figure is what you will try to approximate through the TD prediction algorithm.\n\nYour algorithm for TD prediction has five arguments:\n- `env`: This is an instance of an OpenAI Gym environment.\n- `num_episodes`: This is the number of episodes that are generated through agent-environment interaction.\n- `policy`: This is a 1D numpy array with `policy.shape` equal to the number of states (`env.nS`). `policy[s]` returns the action that the agent chooses when in state `s`.\n- `alpha`: This is the step-size parameter for the update step.\n- `gamma`: This is the discount rate. It must be a value between 0 and 1, inclusive (default value: `1`).\n\nThe algorithm returns as output:\n- `V`: This is a dictionary where `V[s]` is the estimated value of state `s`.\n\nPlease complete the function in the code cell below.", "_____no_output_____" ] ], [ [ "from collections import defaultdict, deque\nimport sys\n\ndef td_prediction(env, num_episodes, policy, alpha, gamma=1.0):\n # initialize empty dictionaries of floats\n V = defaultdict(float)\n # loop over episodes\n for i_episode in range(1, num_episodes+1):\n # monitor progress\n if i_episode % 100 == 0:\n print(\"\\rEpisode {}/{}\".format(i_episode, num_episodes), end=\"\")\n sys.stdout.flush()\n # begin an episode, observe S\n state = env.reset()\n while True:\n # choose action A\n action = policy[state]\n # take action A, observe R, S'\n next_state, reward, done, info = env.step(action)\n # perform updates\n V[state] = V[state] + (alpha * (reward + (gamma * V[next_state]) - V[state])) \n # S <- S'\n state = next_state\n # end episode if reached terminal state\n if done:\n break \n return V", "_____no_output_____" ] ], [ [ "Run the code cell below to test your implementation and visualize the estimated state-value function. If the code cell returns **PASSED**, then you have implemented the function correctly! Feel free to change the `num_episodes` and `alpha` parameters that are supplied to the function. However, if you'd like to ensure the accuracy of the unit test, please do not change the value of `gamma` from the default.", "_____no_output_____" ] ], [ [ "import check_test\n\n# evaluate the policy and reshape the state-value function\nV_pred = td_prediction(env, 5000, policy, .01)\n\n# please do not change the code below this line\nV_pred_plot = np.reshape([V_pred[key] if key in V_pred else 0 for key in np.arange(48)], (4,12)) \ncheck_test.run_check('td_prediction_check', V_pred_plot)\nplot_values(V_pred_plot)", "Episode 5000/5000" ] ], [ [ "How close is your estimated state-value function to the true state-value function corresponding to the policy? \n\nYou might notice that some of the state values are not estimated by the agent. This is because under this policy, the agent will not visit all of the states. In the TD prediction algorithm, the agent can only estimate the values corresponding to states that are visited.", "_____no_output_____" ], [ "### Part 2: TD Control: Sarsa\n\nIn this section, you will write your own implementation of the Sarsa control algorithm.\n\nYour algorithm has four arguments:\n- `env`: This is an instance of an OpenAI Gym environment.\n- `num_episodes`: This is the number of episodes that are generated through agent-environment interaction.\n- `alpha`: This is the step-size parameter for the update step.\n- `gamma`: This is the discount rate. It must be a value between 0 and 1, inclusive (default value: `1`).\n\nThe algorithm returns as output:\n- `Q`: This is a dictionary (of one-dimensional arrays) where `Q[s][a]` is the estimated action value corresponding to state `s` and action `a`.\n\nPlease complete the function in the code cell below.\n\n(_Feel free to define additional functions to help you to organize your code._)", "_____no_output_____" ] ], [ [ "def update_Q(Qsa, Qsa_next, reward, alpha, gamma):\n \"\"\" updates the action-value function estimate using the most recent time step \"\"\"\n return Qsa + (alpha * (reward + (gamma * Qsa_next) - Qsa))\n\ndef epsilon_greedy_probs(env, Q_s, i_episode, eps=None):\n \"\"\" obtains the action probabilities corresponding to epsilon-greedy policy \"\"\"\n epsilon = 1.0 / i_episode\n if eps is not None:\n epsilon = eps\n policy_s = np.ones(env.nA) * epsilon / env.nA\n policy_s[np.argmax(Q_s)] = 1 - epsilon + (epsilon / env.nA)\n return policy_s", "_____no_output_____" ], [ "import matplotlib.pyplot as plt\n%matplotlib inline\n\ndef sarsa(env, num_episodes, alpha, gamma=1.0):\n # initialize action-value function (empty dictionary of arrays)\n Q = defaultdict(lambda: np.zeros(env.nA))\n # initialize performance monitor\n plot_every = 100\n tmp_scores = deque(maxlen=plot_every)\n scores = deque(maxlen=num_episodes)\n # loop over episodes\n for i_episode in range(1, num_episodes+1):\n # monitor progress\n if i_episode % 100 == 0:\n print(\"\\rEpisode {}/{}\".format(i_episode, num_episodes), end=\"\")\n sys.stdout.flush() \n # initialize score\n score = 0\n # begin an episode, observe S\n state = env.reset() \n # get epsilon-greedy action probabilities\n policy_s = epsilon_greedy_probs(env, Q[state], i_episode)\n # pick action A\n action = np.random.choice(np.arange(env.nA), p=policy_s)\n # limit number of time steps per episode\n for t_step in np.arange(300):\n # take action A, observe R, S'\n next_state, reward, done, info = env.step(action)\n # add reward to score\n score += reward\n if not done:\n # get epsilon-greedy action probabilities\n policy_s = epsilon_greedy_probs(env, Q[next_state], i_episode)\n # pick next action A'\n next_action = np.random.choice(np.arange(env.nA), p=policy_s)\n # update TD estimate of Q\n Q[state][action] = update_Q(Q[state][action], Q[next_state][next_action], \n reward, alpha, gamma)\n # S <- S'\n state = next_state\n # A <- A'\n action = next_action\n if done:\n # update TD estimate of Q\n Q[state][action] = update_Q(Q[state][action], 0, reward, alpha, gamma)\n # append score\n tmp_scores.append(score)\n break\n if (i_episode % plot_every == 0):\n scores.append(np.mean(tmp_scores))\n # plot performance\n plt.plot(np.linspace(0,num_episodes,len(scores),endpoint=False),np.asarray(scores))\n plt.xlabel('Episode Number')\n plt.ylabel('Average Reward (Over Next %d Episodes)' % plot_every)\n plt.show()\n # print best 100-episode performance\n print(('Best Average Reward over %d Episodes: ' % plot_every), np.max(scores))\n return Q", "_____no_output_____" ] ], [ [ "Use the next code cell to visualize the **_estimated_** optimal policy and the corresponding state-value function. \n\nIf the code cell returns **PASSED**, then you have implemented the function correctly! Feel free to change the `num_episodes` and `alpha` parameters that are supplied to the function. However, if you'd like to ensure the accuracy of the unit test, please do not change the value of `gamma` from the default.", "_____no_output_____" ] ], [ [ "# obtain the estimated optimal policy and corresponding action-value function\nQ_sarsa = sarsa(env, 5000, .01)\n\n# print the estimated optimal policy\npolicy_sarsa = np.array([np.argmax(Q_sarsa[key]) if key in Q_sarsa else -1 for key in np.arange(48)]).reshape(4,12)\ncheck_test.run_check('td_control_check', policy_sarsa)\nprint(\"\\nEstimated Optimal Policy (UP = 0, RIGHT = 1, DOWN = 2, LEFT = 3, N/A = -1):\")\nprint(policy_sarsa)\n\n# plot the estimated optimal state-value function\nV_sarsa = ([np.max(Q_sarsa[key]) if key in Q_sarsa else 0 for key in np.arange(48)])\nplot_values(V_sarsa)", "Episode 5000/5000" ] ], [ [ "### Part 3: TD Control: Q-learning\n\nIn this section, you will write your own implementation of the Q-learning control algorithm.\n\nYour algorithm has four arguments:\n- `env`: This is an instance of an OpenAI Gym environment.\n- `num_episodes`: This is the number of episodes that are generated through agent-environment interaction.\n- `alpha`: This is the step-size parameter for the update step.\n- `gamma`: This is the discount rate. It must be a value between 0 and 1, inclusive (default value: `1`).\n\nThe algorithm returns as output:\n- `Q`: This is a dictionary (of one-dimensional arrays) where `Q[s][a]` is the estimated action value corresponding to state `s` and action `a`.\n\nPlease complete the function in the code cell below.\n\n(_Feel free to define additional functions to help you to organize your code._)", "_____no_output_____" ] ], [ [ "def q_learning(env, num_episodes, alpha, gamma=1.0):\n # initialize action-value function (empty dictionary of arrays)\n Q = defaultdict(lambda: np.zeros(env.nA))\n # initialize performance monitor\n plot_every = 100\n tmp_scores = deque(maxlen=plot_every)\n scores = deque(maxlen=num_episodes)\n # loop over episodes\n for i_episode in range(1, num_episodes+1):\n # monitor progress\n if i_episode % 100 == 0:\n print(\"\\rEpisode {}/{}\".format(i_episode, num_episodes), end=\"\")\n sys.stdout.flush()\n # initialize score\n score = 0\n # begin an episode, observe S\n state = env.reset()\n while True:\n # get epsilon-greedy action probabilities\n policy_s = epsilon_greedy_probs(env, Q[state], i_episode)\n # pick next action A\n action = np.random.choice(np.arange(env.nA), p=policy_s)\n # take action A, observe R, S'\n next_state, reward, done, info = env.step(action)\n # add reward to score\n score += reward\n # update Q\n Q[state][action] = update_Q(Q[state][action], np.max(Q[next_state]), \\\n reward, alpha, gamma) \n # S <- S'\n state = next_state\n # until S is terminal\n if done:\n # append score\n tmp_scores.append(score)\n break\n if (i_episode % plot_every == 0):\n scores.append(np.mean(tmp_scores))\n # plot performance\n plt.plot(np.linspace(0,num_episodes,len(scores),endpoint=False),np.asarray(scores))\n plt.xlabel('Episode Number')\n plt.ylabel('Average Reward (Over Next %d Episodes)' % plot_every)\n plt.show()\n # print best 100-episode performance\n print(('Best Average Reward over %d Episodes: ' % plot_every), np.max(scores))\n return Q", "_____no_output_____" ] ], [ [ "Use the next code cell to visualize the **_estimated_** optimal policy and the corresponding state-value function. \n\nIf the code cell returns **PASSED**, then you have implemented the function correctly! Feel free to change the `num_episodes` and `alpha` parameters that are supplied to the function. However, if you'd like to ensure the accuracy of the unit test, please do not change the value of `gamma` from the default.", "_____no_output_____" ] ], [ [ "# obtain the estimated optimal policy and corresponding action-value function\nQ_sarsamax = q_learning(env, 5000, .01)\n\n# print the estimated optimal policy\npolicy_sarsamax = np.array([np.argmax(Q_sarsamax[key]) if key in Q_sarsamax else -1 for key in np.arange(48)]).reshape((4,12))\ncheck_test.run_check('td_control_check', policy_sarsamax)\nprint(\"\\nEstimated Optimal Policy (UP = 0, RIGHT = 1, DOWN = 2, LEFT = 3, N/A = -1):\")\nprint(policy_sarsamax)\n\n# plot the estimated optimal state-value function\nplot_values([np.max(Q_sarsamax[key]) if key in Q_sarsamax else 0 for key in np.arange(48)])", "Episode 5000/5000" ] ], [ [ "### Part 4: TD Control: Expected Sarsa\n\nIn this section, you will write your own implementation of the Expected Sarsa control algorithm.\n\nYour algorithm has four arguments:\n- `env`: This is an instance of an OpenAI Gym environment.\n- `num_episodes`: This is the number of episodes that are generated through agent-environment interaction.\n- `alpha`: This is the step-size parameter for the update step.\n- `gamma`: This is the discount rate. It must be a value between 0 and 1, inclusive (default value: `1`).\n\nThe algorithm returns as output:\n- `Q`: This is a dictionary (of one-dimensional arrays) where `Q[s][a]` is the estimated action value corresponding to state `s` and action `a`.\n\nPlease complete the function in the code cell below.\n\n(_Feel free to define additional functions to help you to organize your code._)", "_____no_output_____" ] ], [ [ "def expected_sarsa(env, num_episodes, alpha, gamma=1.0):\n # initialize action-value function (empty dictionary of arrays)\n Q = defaultdict(lambda: np.zeros(env.nA))\n # initialize performance monitor\n plot_every = 100\n tmp_scores = deque(maxlen=plot_every)\n scores = deque(maxlen=num_episodes)\n # loop over episodes\n for i_episode in range(1, num_episodes+1):\n # monitor progress\n if i_episode % 100 == 0:\n print(\"\\rEpisode {}/{}\".format(i_episode, num_episodes), end=\"\")\n sys.stdout.flush()\n # initialize score\n score = 0\n # begin an episode\n state = env.reset()\n # get epsilon-greedy action probabilities\n policy_s = epsilon_greedy_probs(env, Q[state], i_episode, 0.005)\n while True:\n # pick next action\n action = np.random.choice(np.arange(env.nA), p=policy_s)\n # take action A, observe R, S'\n next_state, reward, done, info = env.step(action)\n # add reward to score\n score += reward\n # get epsilon-greedy action probabilities (for S')\n policy_s = epsilon_greedy_probs(env, Q[next_state], i_episode, 0.005)\n # update Q\n Q[state][action] = update_Q(Q[state][action], np.dot(Q[next_state], policy_s), \\\n reward, alpha, gamma) \n # S <- S'\n state = next_state\n # until S is terminal\n if done:\n # append score\n tmp_scores.append(score)\n break\n if (i_episode % plot_every == 0):\n scores.append(np.mean(tmp_scores))\n # plot performance\n plt.plot(np.linspace(0,num_episodes,len(scores),endpoint=False),np.asarray(scores))\n plt.xlabel('Episode Number')\n plt.ylabel('Average Reward (Over Next %d Episodes)' % plot_every)\n plt.show()\n # print best 100-episode performance\n print(('Best Average Reward over %d Episodes: ' % plot_every), np.max(scores))\n return Q", "_____no_output_____" ] ], [ [ "Use the next code cell to visualize the **_estimated_** optimal policy and the corresponding state-value function. \n\nIf the code cell returns **PASSED**, then you have implemented the function correctly! Feel free to change the `num_episodes` and `alpha` parameters that are supplied to the function. However, if you'd like to ensure the accuracy of the unit test, please do not change the value of `gamma` from the default.", "_____no_output_____" ] ], [ [ "# obtain the estimated optimal policy and corresponding action-value function\nQ_expsarsa = expected_sarsa(env, 10000, 1)\n\n# print the estimated optimal policy\npolicy_expsarsa = np.array([np.argmax(Q_expsarsa[key]) if key in Q_expsarsa else -1 for key in np.arange(48)]).reshape(4,12)\ncheck_test.run_check('td_control_check', policy_expsarsa)\nprint(\"\\nEstimated Optimal Policy (UP = 0, RIGHT = 1, DOWN = 2, LEFT = 3, N/A = -1):\")\nprint(policy_expsarsa)\n\n# plot the estimated optimal state-value function\nplot_values([np.max(Q_expsarsa[key]) if key in Q_expsarsa else 0 for key in np.arange(48)])", "Episode 10000/10000" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/elizabethts/DS-Unit-1-Sprint-1-Dealing-With-Data/blob/master/LS_DS_114_Making_Data_backed_Assertions.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "# Lambda School Data Science - Making Data-backed Assertions\n\nThis is, for many, the main point of data science - to create and support reasoned arguments based on evidence. It's not a topic to master in a day, but it is worth some focused time thinking about and structuring your approach to it.", "_____no_output_____" ], [ "## Lecture - generating a confounding variable\n\nThe prewatch material told a story about a hypothetical health condition where both the drug usage and overall health outcome were related to gender - thus making gender a confounding variable, obfuscating the possible relationship between the drug and the outcome.\n\nLet's use Python to generate data that actually behaves in this fashion!", "_____no_output_____" ] ], [ [ "import random\ndir(random) # Reminding ourselves what we can do here\nrandom.seed(10) # Sets Random Seed for Reproducibility", "_____no_output_____" ], [ "# Let's think of another scenario:\n# We work for a company that sells accessories for mobile phones.\n# They have an ecommerce site, and we are supposed to analyze logs\n# to determine what sort of usage is related to purchases, and thus guide\n# website development to encourage higher conversion.\n\n# The hypothesis - users who spend longer on the site tend\n# to spend more. Seems reasonable, no?\n\n# But there's a confounding variable! If they're on a phone, they:\n# a) Spend less time on the site, but\n# b) Are more likely to be interested in the actual products!\n\n# Let's use namedtuple to represent our data\n\nfrom collections import namedtuple\n# purchased and mobile are bools, time_on_site in seconds\nUser = namedtuple('User', ['purchased','time_on_site', 'mobile'])\n\nexample_user = User(False, 12, False)\nprint(example_user)", "User(purchased=False, time_on_site=12, mobile=False)\n" ], [ "# And now let's generate 1000 example users\n# 750 mobile, 250 not (i.e. desktop)\n# A desktop user has a base conversion likelihood of 10%\n# And it goes up by 1% for each 15 seconds they spend on the site\n# And they spend anywhere from 10 seconds to 10 minutes on the site (uniform)\n# Mobile users spend on average half as much time on the site as desktop\n# But have three times as much base likelihood of buying something\n\nusers = []\n\nfor _ in range(250):\n # Desktop users\n time_on_site = random.uniform(10, 600)\n purchased = random.random() < 0.1 + (time_on_site / 1500)\n users.append(User(purchased, time_on_site, False))\n \nfor _ in range(750):\n # Mobile users\n time_on_site = random.uniform(5, 300)\n purchased = random.random() < 0.3 + (time_on_site / 1500)\n users.append(User(purchased, time_on_site, True))\n \nrandom.shuffle(users)\nprint(users[:10])", "[User(purchased=False, time_on_site=172.07500125969045, mobile=False), User(purchased=False, time_on_site=242.1604565076447, mobile=True), User(purchased=True, time_on_site=172.4562884302345, mobile=True), User(purchased=False, time_on_site=134.30741730988564, mobile=True), User(purchased=False, time_on_site=176.6659151415657, mobile=False), User(purchased=False, time_on_site=98.57704667574383, mobile=True), User(purchased=False, time_on_site=141.90635886960914, mobile=True), User(purchased=False, time_on_site=46.30954508769639, mobile=True), User(purchased=True, time_on_site=568.9570603645093, mobile=False), User(purchased=False, time_on_site=64.57737234489078, mobile=True)]\n" ], [ "# Let's put this in a dataframe so we can look at it more easily\nimport pandas as pd\nuser_data = pd.DataFrame(users)\nuser_data.head()", "_____no_output_____" ], [ "user_data.dtypes", "_____no_output_____" ], [ "user_data.isnull().sum()", "_____no_output_____" ], [ "user_data.describe()", "_____no_output_____" ], [ "import numpy as np\nuser_data.describe(exclude=[np.number])", "_____no_output_____" ], [ "# Let's use crosstabulation to try to see what's going on\npd.crosstab(user_data['purchased'], user_data['time_on_site'])", "_____no_output_____" ], [ "# !pip freeze # 0.24.2\n!pip install pandas==0.23.4", "Requirement already satisfied: pandas==0.23.4 in /usr/local/lib/python3.6/dist-packages (0.23.4)\nRequirement already satisfied: pytz>=2011k in /usr/local/lib/python3.6/dist-packages (from pandas==0.23.4) (2018.9)\nRequirement already satisfied: numpy>=1.9.0 in /usr/local/lib/python3.6/dist-packages (from pandas==0.23.4) (1.16.3)\nRequirement already satisfied: python-dateutil>=2.5.0 in /usr/local/lib/python3.6/dist-packages (from pandas==0.23.4) (2.5.3)\nRequirement already satisfied: six>=1.5 in /usr/local/lib/python3.6/dist-packages (from python-dateutil>=2.5.0->pandas==0.23.4) (1.12.0)\n" ], [ "!pip freeze", "absl-py==0.7.1\nalabaster==0.7.12\nalbumentations==0.1.12\naltair==3.0.1\nastor==0.8.0\nastropy==3.0.5\natari-py==0.1.15\natomicwrites==1.3.0\nattrs==19.1.0\naudioread==2.1.7\nautograd==1.2\nBabel==2.6.0\nbackcall==0.1.0\nbackports.tempfile==1.0\nbackports.weakref==1.0.post1\nbeautifulsoup4==4.6.3\nbleach==3.1.0\nblis==0.2.4\nbokeh==1.0.4\nboto==2.49.0\nboto3==1.9.156\nbotocore==1.12.156\nBottleneck==1.2.1\nbranca==0.3.1\nbs4==0.0.1\nbz2file==0.98\ncachetools==3.1.1\ncertifi==2019.3.9\ncffi==1.12.3\nchainer==5.4.0\nchardet==3.0.4\nClick==7.0\ncloudpickle==0.6.1\ncmake==3.12.0\ncolorlover==0.3.0\ncommunity==1.0.0b1\ncontextlib2==0.5.5\nconvertdate==2.1.3\ncoverage==3.7.1\ncoveralls==0.5\ncrcmod==1.7\ncufflinks==0.14.6\ncvxopt==1.2.3\ncvxpy==1.0.15\ncycler==0.10.0\ncymem==2.0.2\nCython==0.29.7\ncytoolz==0.9.0.1\ndaft==0.0.4\ndask==1.1.5\ndataclasses==0.6\ndatascience==0.10.6\ndecorator==4.4.0\ndefusedxml==0.6.0\ndill==0.2.9\ndistributed==1.25.3\nDjango==2.2.1\ndlib==19.16.0\ndm-sonnet==1.32\ndocopt==0.6.2\ndocutils==0.14\ndopamine-rl==1.0.5\neasydict==1.9\necos==2.0.7.post1\neditdistance==0.5.3\nen-core-web-sm==2.0.0\nentrypoints==0.3\nenum34==1.1.6\nephem==3.7.6.0\net-xmlfile==1.0.1\nfa2==0.3.5\nfancyimpute==0.4.3\nfastai==1.0.52\nfastcache==1.1.0\nfastdtw==0.3.2\nfastprogress==0.1.21\nfastrlock==0.4\nfbprophet==0.5\nfeather-format==0.4.0\nfeaturetools==0.4.1\nfilelock==3.0.12\nfix-yahoo-finance==0.0.22\nFlask==1.0.3\nfolium==0.8.3\nfuture==0.16.0\ngast==0.2.2\nGDAL==2.2.2\ngdown==3.6.4\ngensim==3.6.0\ngeographiclib==1.49\ngeopy==1.17.0\ngevent==1.4.0\ngin-config==0.1.4\nglob2==0.6\ngoogle==2.0.2\ngoogle-api-core==1.11.0\ngoogle-api-python-client==1.6.7\ngoogle-auth==1.4.2\ngoogle-auth-httplib2==0.0.3\ngoogle-auth-oauthlib==0.3.0\ngoogle-cloud-bigquery==1.8.1\ngoogle-cloud-core==0.29.1\ngoogle-cloud-language==1.0.2\ngoogle-cloud-storage==1.13.2\ngoogle-cloud-translate==1.3.3\ngoogle-colab==1.0.0\ngoogle-resumable-media==0.3.2\ngoogleapis-common-protos==1.6.0\ngoogledrivedownloader==0.4\ngraph-nets==1.0.4\ngraphviz==0.10.1\ngreenlet==0.4.15\ngrpcio==1.15.0\ngspread==3.0.1\ngspread-dataframe==3.0.2\ngunicorn==19.9.0\ngym==0.10.11\nh5py==2.8.0\nHeapDict==1.0.0\nholidays==0.9.10\nhtml5lib==1.0.1\nhttpimport==0.5.16\nhttplib2==0.11.3\nhumanize==0.5.1\nhyperopt==0.1.2\nideep4py==2.0.0.post3\nidna==2.8\nimage==1.5.27\nimageio==2.4.1\nimagesize==1.1.0\nimbalanced-learn==0.4.3\nimblearn==0.0\nimgaug==0.2.9\nimutils==0.5.2\ninflect==2.1.0\nintel-openmp==2019.0\nintervaltree==2.1.0\nipykernel==4.6.1\nipython==5.5.0\nipython-genutils==0.2.0\nipython-sql==0.3.9\nipywidgets==7.4.2\nitsdangerous==1.1.0\njdcal==1.4.1\njedi==0.13.3\njieba==0.39\nJinja2==2.10.1\njmespath==0.9.4\njoblib==0.12.5\njpeg4py==0.1.4\njsonschema==2.6.0\njupyter==1.0.0\njupyter-client==5.2.4\njupyter-console==6.0.0\njupyter-core==4.4.0\nkaggle==1.5.3\nkapre==0.1.3.1\nKeras==2.2.4\nKeras-Applications==1.0.7\nKeras-Preprocessing==1.0.9\nkeras-vis==0.4.1\nkiwisolver==1.1.0\nknnimpute==0.1.0\nlibrosa==0.6.3\nlightgbm==2.2.3\nllvmlite==0.28.0\nlmdb==0.94\nlucid==0.3.8\nlunardate==0.2.0\nlxml==4.2.6\nmagenta==0.3.19\nMarkdown==3.1.1\nMarkupSafe==1.1.1\nmatplotlib==3.0.3\nmatplotlib-venn==0.11.5\nmesh-tensorflow==0.0.5\nmido==1.2.6\nmir-eval==0.5\nmissingno==0.4.1\nmistune==0.8.4\nmkl==2019.0\nmlxtend==0.14.0\nmock==3.0.5\nmore-itertools==7.0.0\nmoviepy==0.2.3.5\nmpi4py==3.0.1\nmpmath==1.1.0\nmsgpack==0.5.6\nmsgpack-numpy==0.4.3.2\nmultiprocess==0.70.7\nmultitasking==0.0.9\nmurmurhash==1.0.2\nmusic21==5.5.0\nnatsort==5.5.0\nnbconvert==5.5.0\nnbformat==4.4.0\nnetworkx==2.3\nnibabel==2.3.3\nnltk==3.2.5\nnose==1.3.7\nnotebook==5.2.2\nnp-utils==0.5.10.0\nnumba==0.40.1\nnumexpr==2.6.9\nnumpy==1.16.3\nnvidia-ml-py3==7.352.0\noauth2client==4.1.3\noauthlib==3.0.1\nokgrade==0.4.3\nolefile==0.46\nopencv-contrib-python==3.4.3.18\nopencv-python==3.4.5.20\nopenpyxl==2.5.9\nosqp==0.5.0\npackaging==19.0\npandas==0.23.4\npandas-datareader==0.7.0\npandas-gbq==0.4.1\npandas-profiling==1.4.1\npandocfilters==1.4.2\nparso==0.4.0\npathlib==1.0.1\npatsy==0.5.1\npexpect==4.7.0\npickleshare==0.7.5\nPillow==4.3.0\npip-tools==3.6.1\nplac==0.9.6\nplotly==3.6.1\npluggy==0.7.1\nportpicker==1.2.0\nprefetch-generator==1.0.1\npreshed==2.0.1\npretty-midi==0.2.8\nprettytable==0.7.2\nprogressbar2==3.38.0\nprometheus-client==0.6.0\npromise==2.2.1\nprompt-toolkit==1.0.16\nprotobuf==3.7.1\npsutil==5.4.8\npsycopg2==2.7.6.1\nptyprocess==0.6.0\npy==1.8.0\npyarrow==0.13.0\npyasn1==0.4.5\npyasn1-modules==0.2.5\npycocotools==2.0.0\npycparser==2.19\npydot==1.3.0\npydot-ng==2.0.0\npydotplus==2.0.2\npyemd==0.5.1\npyglet==1.3.2\nPygments==2.1.3\npygobject==3.26.1\npymc3==3.6\npymongo==3.8.0\npymystem3==0.2.0\nPyOpenGL==3.1.0\npyparsing==2.4.0\npyrsistent==0.15.2\npysndfile==1.3.2\nPySocks==1.7.0\npystan==2.19.0.0\npytest==3.6.4\npython-apt==1.6.4\npython-chess==0.23.11\npython-dateutil==2.5.3\npython-louvain==0.13\npython-rtmidi==1.3.0\npython-slugify==3.0.2\npython-utils==2.3.0\npytz==2018.9\nPyWavelets==1.0.3\nPyYAML==3.13\npyzmq==17.0.0\nqtconsole==4.5.1\nregex==2018.1.10\nrequests==2.21.0\nrequests-oauthlib==1.2.0\nresampy==0.2.1\nretrying==1.3.3\nrpy2==2.9.5\nrsa==4.0\ns3fs==0.2.1\ns3transfer==0.2.0\nscikit-image==0.15.0\nscikit-learn==0.21.2\nscipy==1.3.0\nscreen-resolution-extra==0.0.0\nscs==2.1.0\nseaborn==0.9.0\nsemantic-version==2.6.0\nSend2Trash==1.5.0\nsetuptools-git==1.2\nShapely==1.6.4.post2\nsimplegeneric==0.8.1\nsix==1.12.0\nsklearn==0.0\nsklearn-pandas==1.8.0\nsmart-open==1.8.3\nsnowballstemmer==1.2.1\nsortedcontainers==2.1.0\nsoupsieve==1.9.1\nspacy==2.0.18\nSphinx==1.8.5\nsphinxcontrib-websupport==1.1.2\nSQLAlchemy==1.3.3\nsqlparse==0.3.0\nsrsly==0.0.5\nstable-baselines==2.2.1\nstatsmodels==0.9.0\nsympy==1.1.1\ntables==3.4.4\ntabulate==0.8.3\ntblib==1.4.0\ntensor2tensor==1.11.0\ntensorboard==1.13.1\ntensorboardcolab==0.0.22\ntensorflow==1.13.1\ntensorflow-estimator==1.13.0\ntensorflow-hub==0.4.0\ntensorflow-metadata==0.13.0\ntensorflow-probability==0.6.0\ntermcolor==1.1.0\nterminado==0.8.2\ntestpath==0.4.2\ntext-unidecode==1.2\ntextblob==0.15.3\ntextgenrnn==1.4.1\ntfds-nightly==1.0.2.dev201905240105\ntflearn==0.3.2\nTheano==1.0.4\nthinc==6.12.1\ntoolz==0.9.0\ntorch==1.1.0\ntorchsummary==1.5.1\ntorchtext==0.3.1\ntorchvision==0.3.0\ntornado==4.5.3\ntqdm==4.28.1\ntraitlets==4.3.2\ntweepy==3.6.0\ntyping==3.6.6\ntzlocal==1.5.1\nujson==1.35\numap-learn==0.3.8\nuritemplate==3.0.0\nurllib3==1.24.3\nvega-datasets==0.7.0\nwasabi==0.2.2\nwcwidth==0.1.7\nwebencodings==0.5.1\nWerkzeug==0.15.4\nwidgetsnbextension==3.4.2\nwordcloud==1.5.0\nwrapt==1.10.11\nxarray==0.11.3\nxgboost==0.90\nxkit==0.0.0\nxlrd==1.1.0\nxlwt==1.3.0\nyellowbrick==0.9.1\nzict==0.1.4\nzmq==0.0.0\n" ], [ "# OK, that's not quite what we want\n# Time is continuous! We need to put it in discrete buckets\n# Pandas calls these bins, and pandas.cut helps make them\n\ntime_bins = pd.cut(user_data['time_on_site'], 6) # 6 equal-sized bins\npd.crosstab(user_data['purchased'], time_bins)", "_____no_output_____" ], [ "# We can make this a bit clearer by normalizing (getting %)\npd.crosstab(user_data['purchased'], time_bins, normalize='columns')", "_____no_output_____" ], [ "# That seems counter to our hypothesis\n# More time on the site can actually have fewer purchases\n\n# But we know why, since we generated the data!\n# Let's look at mobile and purchased\npd.crosstab(user_data['purchased'], user_data['mobile'], normalize='columns')", "_____no_output_____" ], [ "# Yep, mobile users are more likely to buy things\n# But we're still not seeing the *whole* story until we look at all 3 at once\n\n# Live/stretch goal - how can we do that?\n\nct = pd.crosstab(user_data['mobile'], [user_data['purchased'], time_bins],\n rownames = ['device'],\n colnames = ['purchased','time_on_site'],\n normalize = 'index'\n )\nct\n", "_____no_output_____" ], [ "type(ct)", "_____no_output_____" ], [ "pt = pd.pivot_table(user_data, values = 'purchased', index = time_bins)\npt", "_____no_output_____" ], [ "pt.plot.bar();", "_____no_output_____" ], [ "ct = pd.crosstab(time_bins, [user_data['purchased'], user_data['mobile']], \n normalize = 'columns')\nct", "_____no_output_____" ], [ "ct_final = ct.iloc[:, [2,3]]\nct_final.plot(kind = 'bar', stacked = True);", "_____no_output_____" ] ], [ [ "## Assignment - what's going on here?\n\nConsider the data in `persons.csv` (already prepared for you, in the repo for the week). It has four columns - a unique id, followed by age (in years), weight (in lbs), and exercise time (in minutes/week) of 1200 (hypothetical) people.\n\nTry to figure out which variables are possibly related to each other, and which may be confounding relationships.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt", "_____no_output_____" ], [ "# TODO - your code here\n# Use what we did live in lecture as an example\n\n# HINT - you can find the raw URL on GitHub and potentially use that\n# to load the data with read_csv, or you can upload it yourself\n\n\n#load data\npersons_url = 'https://raw.githubusercontent.com/LambdaSchool/DS-Unit-1-Sprint-1-Dealing-With-Data/master/module3-databackedassertions/persons.csv'\npersons_df = pd.read_csv(persons_url)\npersons_df.head()", "_____no_output_____" ], [ "persons_df.shape ", "_____no_output_____" ], [ "time_range = pd.cut(persons_df['exercise_time'],30) #cut time into 5 groups\nage_range = pd.cut(persons_df['age'],3) #cut time into 5 groups\nweight_range = pd.cut(persons_df['weight'],30)\npersons_ct = pd.crosstab(persons_df['weight'], \n [age_range, persons_df['exercise_time']])\npersons_ct", "_____no_output_____" ], [ "group1 = persons_df[persons_df.age <30]\ngroup2 = persons_df[(persons_df.age >=30) & (persons_df.age<50)]\ngroup3 = persons_df[persons_df.age >=50]", "_____no_output_____" ], [ "fig, ax1 = plt.subplots()\nax1.set_axisbelow(True)\nax1.set_xlabel('Exercise Time')\nax1.set_ylabel('Weight')\nax1.set_title('Weight vs Exercise Time (Age <30)')\nplt.gca().invert_yaxis()\nplt.scatter(group1.exercise_time, group1.weight, color ='red', alpha = .4)\n\nfig, ax2 = plt.subplots()\nax2.set_axisbelow(True)\nax2.set_xlabel('Exercise Time')\nax2.set_ylabel('Weight')\nax2.set_title('Weight vs Exercise Time (Age 30-50)')\nplt.gca().invert_yaxis()\nplt.scatter(group2.exercise_time, group2.weight, color = 'teal', alpha = .4)\n\nfig, ax3 = plt.subplots()\nax3.set_axisbelow(True)\nax3.set_xlabel('Exercise Time')\nax3.set_ylabel('Weight')\nax3.set_title('Weight vs Exercise Time (Age > 50)')\nplt.gca().invert_yaxis()\nplt.scatter(group3.exercise_time, group3.weight, color = 'yellow', alpha = 0.4);\n\n\nfig, ax4 = plt.subplots()\nax4.set_axisbelow(True)\nax4.set_xlabel('Exercise Time')\nax4.set_ylabel('Weight')\nax4.set_title('Weight vs Exercise Time for All Ages')\nplt.scatter(group2.exercise_time, group2.weight, color = 'teal', alpha = .6)\nplt.scatter(group3.exercise_time, group3.weight,color = 'yellow', alpha = 0.6);\nplt.scatter(group1.exercise_time, group1.weight, color ='red', alpha = .6)\nplt.gca().invert_yaxis()\n\n\nfig, ax5 = plt.subplots()\nax5.set_axisbelow(True)\nax5.set_xlabel('Exercise Time')\nax5.set_ylabel('Age')\nax5.set_title('Age vs Exercise Time (All Ages)')\nplt.scatter(group2.exercise_time, group2.age, color = 'teal', alpha = .6)\nplt.scatter(group3.exercise_time, group3.age,color = 'yellow', alpha = 0.6);\nplt.scatter(group1.exercise_time, group1.age, color ='red', alpha = .6)\n\nfig, ax6 = plt.subplots()\nax6.set_axisbelow(True)\nax6.set_xlabel('Weight')\nax6.set_ylabel('Age')\nax6.set_title('Age vs Weight (All Ages)')\nplt.scatter(group2.weight, group2.age, color = 'teal', alpha = .6)\nplt.scatter(group3.weight, group3.age,color = 'yellow', alpha = 0.6);\nplt.scatter(group1.weight, group1.age, color ='red', alpha = .6);\n", "_____no_output_____" ], [ "#Conclusions: \n# -People who exercise more weight less\n# -The group age >=50 tends to exercise for a shorter amount of time ", "_____no_output_____" ] ], [ [ "### Assignment questions\n\nAfter you've worked on some code, answer the following questions in this text block:\n\n1. What are the variable types in the data?\n \n \n age - ordinal and continuous\n weight - ordinal and continuous\n exercise time - ordinal and continuous\n\n2. What are the relationships between the variables?\n \n \n Weight decreases as exercise time increases so they are inversely related. Weight increases as age increases so they are directly related\n\n3. Which relationships are \"real\", and which spurious?\n\n\n The relationship between weight and exercise time is real and the relationship between age and weight is spurious (older people tend to exercise less)\n", "_____no_output_____" ], [ "## Stretch goals and resources\n\nFollowing are *optional* things for you to take a look at. Focus on the above assignment first, and make sure to commit and push your changes to GitHub.\n\n- [Spurious Correlations](http://tylervigen.com/spurious-correlations)\n- [NIH on controlling for confounding variables](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4017459/)\n\nStretch goals:\n\n- Produce your own plot inspired by the Spurious Correlation visualizations (and consider writing a blog post about it - both the content and how you made it)\n- Pick one of the techniques that NIH highlights for confounding variables - we'll be going into many of them later, but see if you can find which Python modules may help (hint - check scikit-learn)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n", "_____no_output_____" ], [ "#data importation\ndf = pd.read_csv(\"HR-Employee-Attrition.csv\")\n\ndf.head()", "_____no_output_____" ], [ "#data analysis\ndf.dtypes", "_____no_output_____" ], [ "#data analysis\ndf.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 1470 entries, 0 to 1469\nData columns (total 35 columns):\n # Column Non-Null Count Dtype \n--- ------ -------------- ----- \n 0 Age 1470 non-null int64 \n 1 Attrition 1470 non-null object\n 2 BusinessTravel 1470 non-null object\n 3 DailyRate 1470 non-null int64 \n 4 Department 1470 non-null object\n 5 DistanceFromHome 1470 non-null int64 \n 6 Education 1470 non-null int64 \n 7 EducationField 1470 non-null object\n 8 EmployeeCount 1470 non-null int64 \n 9 EmployeeNumber 1470 non-null int64 \n 10 EnvironmentSatisfaction 1470 non-null int64 \n 11 Gender 1470 non-null object\n 12 HourlyRate 1470 non-null int64 \n 13 JobInvolvement 1470 non-null int64 \n 14 JobLevel 1470 non-null int64 \n 15 JobRole 1470 non-null object\n 16 JobSatisfaction 1470 non-null int64 \n 17 MaritalStatus 1470 non-null object\n 18 MonthlyIncome 1470 non-null int64 \n 19 MonthlyRate 1470 non-null int64 \n 20 NumCompaniesWorked 1470 non-null int64 \n 21 Over18 1470 non-null object\n 22 OverTime 1470 non-null object\n 23 PercentSalaryHike 1470 non-null int64 \n 24 PerformanceRating 1470 non-null int64 \n 25 RelationshipSatisfaction 1470 non-null int64 \n 26 StandardHours 1470 non-null int64 \n 27 StockOptionLevel 1470 non-null int64 \n 28 TotalWorkingYears 1470 non-null int64 \n 29 TrainingTimesLastYear 1470 non-null int64 \n 30 WorkLifeBalance 1470 non-null int64 \n 31 YearsAtCompany 1470 non-null int64 \n 32 YearsInCurrentRole 1470 non-null int64 \n 33 YearsSinceLastPromotion 1470 non-null int64 \n 34 YearsWithCurrManager 1470 non-null int64 \ndtypes: int64(26), object(9)\nmemory usage: 402.1+ KB\n" ], [ "#Showing a breakdown of distance from home by job role and attrition.\ndf.groupby([\"DistanceFromHome\", \"JobRole\"])[\"Attrition\"].count()\n\n\n", "_____no_output_____" ], [ "#Comparing average monthly income by education and attrition.\ndf.groupby([\"MonthlyIncome\", \"Education\"])[\"Attrition\"].count()\n\n\n\n\n", "_____no_output_____" ], [ "#getting average montly income income and creating column\nav_Monthlyincome = df[\"MonthlyIncome\"].mean()\n\ndf = pd.DataFrame(df)\ndf['av_Monthlyincome'] = av_Monthlyincome\n\ndf.head()\n\n", "_____no_output_____" ], [ "# comparing the three columns\ndf_comparison = pd.DataFrame(df, columns = ['av_Monthlyincome','Education', 'Attrition'] )\n\ndf_comparison.head(10)\n", "_____no_output_____" ], [ "#ASSIGHNMENT 3\n#Based on your analysis in Assignment 2, express your results in the\n#form of the most appropriate visualization using any visualization library of your choice.", "_____no_output_____" ], [ "# visualisation on breakdown of distance from home by job role and attrition.\n\nsns.catplot(x=\"DistanceFromHome\",y=\"JobRole\",hue=\"Attrition\",data=df,palette=\"Set2\")\nplt.title(\"How Distance from home and job role effect the attrition\");\n\n\n", "_____no_output_____" ], [ "# visualisation of Comparing average monthly income by education and attrition\n\n#fig, ax = plt.subplots(figsize = (8,8))\n#sns.barplot(y = 'av_Monthlyincome',x = 'Education',hue = 'Attrition',data = df,ax = ax)\n", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%load_ext autoreload\n%autoreload 2\n%matplotlib inline\nimport numpy as np\nimport torch\nimport matplotlib.pyplot as plt\nfrom tqdm import tqdm\nimport torch.nn as nn\nimport torch.nn.functional as F\nimport torch.optim as optim\nimport util\nimport sys\nsys.path.append(\"../..\")", "The autoreload extension is already loaded. To reload it, use:\n %reload_ext autoreload\n" ], [ "import cifar10\ndevice = 'cuda' if torch.cuda.is_available() else 'cpu'\nprint(device)\n\n# load data \ndata_path = \"../../data/cifar10\"\ntrain_loader, test_loader = cifar10.get_dataloader(data_path, batch_size=100, download=True)", "cuda\nFiles already downloaded and verified\nFiles already downloaded and verified\n" ], [ "class Net(nn.Module):\n def __init__(self):\n super().__init__()\n self.conv1 = nn.Conv2d(3, 6, 5)\n self.pool = nn.MaxPool2d(2, 2)\n self.conv2 = nn.Conv2d(6, 16, 5)\n self.fc1 = nn.Linear(16 * 5 * 5, 120)\n self.fc2 = nn.Linear(120, 84)\n self.fc3 = nn.Linear(84, 10)\n\n def forward(self, x):\n x = self.pool(F.relu(self.conv1(x)))\n x = self.pool(F.relu(self.conv2(x)))\n x = torch.flatten(x, 1) # flatten all dimensions except batch\n x = F.relu(self.fc1(x))\n x = F.relu(self.fc2(x))\n x = self.fc3(x)\n return x\n \n# define model and hyperparms\ncriterion = nn.CrossEntropyLoss()\noptimizer = optim.Adam(net.parameters(), lr=0.001)\nnum_epochs = 10\nnet = Net().to(device) ", "_____no_output_____" ] ], [ [ "# train cnn", "_____no_output_____" ] ], [ [ "util.train(model, device, train_loader, test_loader, optimizer, num_epochs, criterion)", "Train Epoch: 1 [45000/50000 (90%)]\tLoss: 1.500776\nTest set: Average loss: 0.0002, Accuracy: 4550/10000 (45.50%)\n\nTrain Epoch: 2 [45000/50000 (90%)]\tLoss: 1.379729\nTest set: Average loss: 0.0001, Accuracy: 5390/10000 (53.90%)\n\nTrain Epoch: 3 [45000/50000 (90%)]\tLoss: 1.330050\nTest set: Average loss: 0.0001, Accuracy: 5640/10000 (56.40%)\n\nTrain Epoch: 4 [45000/50000 (90%)]\tLoss: 1.189881\nTest set: Average loss: 0.0001, Accuracy: 5753/10000 (57.53%)\n\nTrain Epoch: 5 [45000/50000 (90%)]\tLoss: 1.287006\nTest set: Average loss: 0.0001, Accuracy: 6042/10000 (60.42%)\n\nTrain Epoch: 6 [45000/50000 (90%)]\tLoss: 1.345478\nTest set: Average loss: 0.0001, Accuracy: 5962/10000 (59.62%)\n\nTrain Epoch: 7 [45000/50000 (90%)]\tLoss: 1.042157\nTest set: Average loss: 0.0001, Accuracy: 6065/10000 (60.65%)\n\nTrain Epoch: 8 [45000/50000 (90%)]\tLoss: 1.040688\nTest set: Average loss: 0.0001, Accuracy: 6236/10000 (62.36%)\n\nTrain Epoch: 9 [45000/50000 (90%)]\tLoss: 1.040002\nTest set: Average loss: 0.0001, Accuracy: 6186/10000 (61.86%)\n\nTrain Epoch: 10 [45000/50000 (90%)]\tLoss: 0.792405\nTest set: Average loss: 0.0001, Accuracy: 6311/10000 (63.11%)\n\n" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import numpy as np\nimport pandas_datareader as pdr\nimport datetime as dt\nimport pandas as pd\nimport matplotlib.pyplot as plt", "/usr/lib/python3/dist-packages/requests/__init__.py:89: RequestsDependencyWarning: urllib3 (1.26.7) or chardet (3.0.4) doesn't match a supported version!\n warnings.warn(\"urllib3 ({}) or chardet ({}) doesn't match a supported \"\n" ], [ "start = dt.datetime(2012, 6, 1)\nend = dt.datetime(2022, 3, 9)\n\nstock = ['fb']\nstock_data = pdr.get_data_yahoo(stock, start, end)\n\npair = ['snap']\npair_data = pdr.get_data_yahoo(pair, start, end)\n\n", "_____no_output_____" ], [ "main_df = pd.DataFrame()\npair_df = pd.DataFrame()\n\nmain_df[\"Close\"] = stock_data[[\"Close\"]]\nmain_df[\"Open\"] = stock_data[[\"Open\"]]\nmain_df[\"High\"] = stock_data[[\"High\"]]\nmain_df[\"Low\"] = stock_data[[\"Low\"]]\nmain_df[\"Volume\"] = stock_data[[\"Volume\"]]\nmain_df[\"stock_prev_close\"] = main_df[\"Close\"].shift(1)\nmain_df['Date'] = main_df.index\n\npair_df[\"Close\"] = pair_data[[\"Close\"]]\npair_df[\"Open\"] = pair_data[[\"Open\"]]\npair_df[\"High\"] = pair_data[[\"High\"]]\npair_df[\"Low\"] = pair_data[[\"Low\"]]\npair_df[\"Volume\"] = pair_data[[\"Volume\"]]\npair_df[\"pair_prev_close\"] = pair_df[\"Close\"].shift(1)\npair_df['Date'] = pair_df.index\n", "_____no_output_____" ], [ "merged = pd.merge(main_df.reset_index(drop=True), pair_df.reset_index(drop=True), on=[\"Date\"], how=\"left\")", "_____no_output_____" ], [ "merged", "_____no_output_____" ], [ "merged['ratio_spread'] = merged['Close_x'] / merged['Close_y']\nmerged['prev_ratio_spread'] = merged['ratio_spread'].shift(1)\nmerged['close_spread'] = np.log(merged['Close_x']) - np.log(merged['Close_y'])\nmerged['close_spread_sma20'] = merged['close_spread'].rolling(20).mean()\nmerged['close_spread_std20'] = merged['close_spread'].rolling(20).std()\nmerged['zscore'] = (merged['close_spread'] - merged['close_spread_sma20']) / merged['close_spread_std20']\nmerged['prev_zscore'] = merged['zscore'].shift(1)\nmerged['lsr'] = np.log(merged['ratio_spread']) - np.log(merged['prev_ratio_spread'])\nmerged['strat_return'] = merged['prev_zscore'] * -1 * (merged['lsr'] - np.log(merged['ratio_spread']))", "_____no_output_____" ], [ "merged", "_____no_output_____" ], [ "figsize=(10,5)\nax = merged.plot(x=\"Date\", y=\"zscore\", legend=False,figsize=figsize)", "_____no_output_____" ], [ "figsize=(10,5)\nax = merged.plot(x=\"Date\", y=\"lsr\", legend=False,figsize=figsize)", "_____no_output_____" ], [ "figsize=(10,5)\nax = merged.plot(x=\"Date\", y=\"strat_return\", legend=False,figsize=figsize)", "_____no_output_____" ], [ "figsize=(30,10)\nax = merged[1200:].plot(y=\"strat_return\", legend=False,figsize=figsize)\n", "_____no_output_____" ] ], [ [ "# Monthly Aggregates - Work in Progress - I forgot to set price to 0 - Need to rewatch video!", "_____no_output_____" ] ], [ [ "merged['date_trunc_month'] = merged.apply(lambda x: dt.datetime.strftime(x['Date'],'%Y-%m'), axis=1)", "_____no_output_____" ], [ "merged['date_trunc_month']", "_____no_output_____" ], [ "merged['is_start_of_month'] = merged.apply(lambda x: x.name == min(merged[merged['date_trunc_month'] == x['date_trunc_month']].index), axis=1)\nmerged['is_end_of_month'] = merged.apply(lambda x: x.name == max(merged[merged['date_trunc_month'] == x['date_trunc_month']].index), axis=1)", "_____no_output_____" ], [ "merged[merged['is_start_of_month'] == True]", "_____no_output_____" ] ], [ [ "```sql\nSELECT\n startofmonth,\n stock1,\n stock2,\n SUM(lsr) as lsr\nFROM dailyreturns\nGROUP BY\n startofmonth,\n stock1,\n stock2\n```", "_____no_output_____" ] ], [ [ "merged[(merged['is_start_of_month'] == True)][['date_trunc_month','lsr']]", "_____no_output_____" ], [ "merged[(merged['is_start_of_month'] == True) | (merged['is_end_of_month'] == True)][['date_trunc_month','lsr']].plot()", "_____no_output_____" ], [ "merged['lsr'] = merged.apply(lambda x: 0 if x['is_end_of_month'] is True else x['lsr'], axis=1)", "_____no_output_____" ], [ "merged[merged['is_end_of_month'] == True]", "_____no_output_____" ], [ "monthly_agg = merged[(merged['is_start_of_month'] == True) | (merged['is_end_of_month'] == True)].copy()", "_____no_output_____" ], [ "monthly_agg.reset_index(drop=True)\nmonthly_agg = monthly_agg.dropna()", "_____no_output_____" ], [ "monthly_agg", "_____no_output_____" ], [ "figsize=(30,10)\nax = monthly_agg.plot(y=\"lsr\", x=\"date_trunc_month\", legend=False,figsize=figsize)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Convolutional Neural Networks\n---\nIn this notebook, we train a **CNN** to classify images from the CIFAR-10 database.\n\nThe images in this database are small color images that fall into one of ten classes; some example images are pictured below.\n\n<img src='notebook_ims/cifar_data.png' width=70% height=70% />", "_____no_output_____" ], [ "### Test for [CUDA](http://pytorch.org/docs/stable/cuda.html)\n\nSince these are larger (32x32x3) images, it may prove useful to speed up your training time by using a GPU. CUDA is a parallel computing platform and CUDA Tensors are the same as typical Tensors, only they utilize GPU's for computation.", "_____no_output_____" ] ], [ [ "import torch\nimport numpy as np\n\n# check if CUDA is available\ntrain_on_gpu = torch.cuda.is_available()\n\nif not train_on_gpu:\n print('CUDA is not available. Training on CPU ...')\nelse:\n print('CUDA is available! Training on GPU ...')", "CUDA is available! Training on GPU ...\n" ] ], [ [ "---\n## Load and Augment the [Data](http://pytorch.org/docs/stable/torchvision/datasets.html)\n\nDownloading may take a minute. We load in the training and test data, split the training data into a training and validation set, then create DataLoaders for each of these sets of data.\n\n#### Augmentation\n\nIn this cell, we perform some simple [data augmentation](https://medium.com/nanonets/how-to-use-deep-learning-when-you-have-limited-data-part-2-data-augmentation-c26971dc8ced) by randomly flipping and rotating the given image data. We do this by defining a torchvision `transform`, and you can learn about all the transforms that are used to pre-process and augment data, [here](https://pytorch.org/docs/stable/torchvision/transforms.html).\n\n#### TODO: Look at the [transformation documentation](https://pytorch.org/docs/stable/torchvision/transforms.html); add more augmentation transforms, and see how your model performs.\n\nThis type of data augmentation should add some positional variety to these images, so that when we train a model on this data, it will be robust in the face of geometric changes (i.e. it will recognize a ship, no matter which direction it is facing). It's recommended that you choose one or two transforms.", "_____no_output_____" ] ], [ [ "from torchvision import datasets\nimport torchvision.transforms as transforms\nfrom torch.utils.data.sampler import SubsetRandomSampler\n\n# number of subprocesses to use for data loading\nnum_workers = 0\n# how many samples per batch to load\nbatch_size = 20\n# percentage of training set to use as validation\nvalid_size = 0.2\n\n# convert data to a normalized torch.FloatTensor\ntransform = transforms.Compose([\n transforms.RandomHorizontalFlip(), # randomly flip and rotate\n transforms.RandomRotation(10),\n transforms.ToTensor(),\n transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))\n ])\n\n# choose the training and test datasets\ntrain_data = datasets.CIFAR10('data', train=True,\n download=True, transform=transform)\ntest_data = datasets.CIFAR10('data', train=False,\n download=True, transform=transform)\n\n# obtain training indices that will be used for validation\nnum_train = len(train_data)\nindices = list(range(num_train))\nnp.random.shuffle(indices)\nsplit = int(np.floor(valid_size * num_train))\ntrain_idx, valid_idx = indices[split:], indices[:split]\n\n# define samplers for obtaining training and validation batches\ntrain_sampler = SubsetRandomSampler(train_idx)\nvalid_sampler = SubsetRandomSampler(valid_idx)\n\n# prepare data loaders (combine dataset and sampler)\ntrain_loader = torch.utils.data.DataLoader(train_data, batch_size=batch_size,\n sampler=train_sampler, num_workers=num_workers)\nvalid_loader = torch.utils.data.DataLoader(train_data, batch_size=batch_size, \n sampler=valid_sampler, num_workers=num_workers)\ntest_loader = torch.utils.data.DataLoader(test_data, batch_size=batch_size, \n num_workers=num_workers)\n\n# specify the image classes\nclasses = ['airplane', 'automobile', 'bird', 'cat', 'deer',\n 'dog', 'frog', 'horse', 'ship', 'truck']", "Files already downloaded and verified\nFiles already downloaded and verified\n" ] ], [ [ "### Visualize a Batch of Training Data", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\n%matplotlib inline\n\n# helper function to un-normalize and display an image\ndef imshow(img):\n img = img / 2 + 0.5 # unnormalize\n plt.imshow(np.transpose(img, (1, 2, 0))) # convert from Tensor image", "_____no_output_____" ], [ "# obtain one batch of training images\ndataiter = iter(train_loader)\nimages, labels = dataiter.next()\nimages = images.numpy() # convert images to numpy for display\n\n# plot the images in the batch, along with the corresponding labels\nfig = plt.figure(figsize=(25, 4))\n# display 20 images\nfor idx in np.arange(20):\n ax = fig.add_subplot(2, 20/2, idx+1, xticks=[], yticks=[])\n imshow(images[idx])\n ax.set_title(classes[labels[idx]])", "_____no_output_____" ] ], [ [ "### View an Image in More Detail\n\nHere, we look at the normalized red, green, and blue (RGB) color channels as three separate, grayscale intensity images.", "_____no_output_____" ] ], [ [ "rgb_img = np.squeeze(images[3])\nchannels = ['red channel', 'green channel', 'blue channel']\n\nfig = plt.figure(figsize = (36, 36)) \nfor idx in np.arange(rgb_img.shape[0]):\n ax = fig.add_subplot(1, 3, idx + 1)\n img = rgb_img[idx]\n ax.imshow(img, cmap='gray')\n ax.set_title(channels[idx])\n width, height = img.shape\n thresh = img.max()/2.5\n for x in range(width):\n for y in range(height):\n val = round(img[x][y],2) if img[x][y] !=0 else 0\n ax.annotate(str(val), xy=(y,x),\n horizontalalignment='center',\n verticalalignment='center', size=8,\n color='white' if img[x][y]<thresh else 'black')", "_____no_output_____" ] ], [ [ "---\n## Define the Network [Architecture](http://pytorch.org/docs/stable/nn.html)\n\nThis time, you'll define a CNN architecture. Instead of an MLP, which used linear, fully-connected layers, you'll use the following:\n* [Convolutional layers](https://pytorch.org/docs/stable/nn.html#conv2d), which can be thought of as stack of filtered images.\n* [Maxpooling layers](https://pytorch.org/docs/stable/nn.html#maxpool2d), which reduce the x-y size of an input, keeping only the most _active_ pixels from the previous layer.\n* The usual Linear + Dropout layers to avoid overfitting and produce a 10-dim output.\n\nA network with 2 convolutional layers is shown in the image below and in the code, and you've been given starter code with one convolutional and one maxpooling layer.\n\n<img src='notebook_ims/2_layer_conv.png' height=50% width=50% />\n\n#### TODO: Define a model with multiple convolutional layers, and define the feedforward metwork behavior.\n\nThe more convolutional layers you include, the more complex patterns in color and shape a model can detect. It's suggested that your final model include 2 or 3 convolutional layers as well as linear layers + dropout in between to avoid overfitting. \n\nIt's good practice to look at existing research and implementations of related models as a starting point for defining your own models. You may find it useful to look at [this PyTorch classification example](https://github.com/pytorch/tutorials/blob/master/beginner_source/blitz/cifar10_tutorial.py) or [this, more complex Keras example](https://github.com/keras-team/keras/blob/master/examples/cifar10_cnn.py) to help decide on a final structure.\n\n#### Output volume for a convolutional layer\n\nTo compute the output size of a given convolutional layer we can perform the following calculation (taken from [Stanford's cs231n course](http://cs231n.github.io/convolutional-networks/#layers)):\n> We can compute the spatial size of the output volume as a function of the input volume size (W), the kernel/filter size (F), the stride with which they are applied (S), and the amount of zero padding used (P) on the border. The correct formula for calculating how many neurons define the output_W is given by `(W−F+2P)/S+1`. \n\nFor example for a 7x7 input and a 3x3 filter with stride 1 and pad 0 we would get a 5x5 output. With stride 2 we would get a 3x3 output.", "_____no_output_____" ] ], [ [ "import torch.nn as nn\nimport torch.nn.functional as F\n\n# define the CNN architecture\nclass Net(nn.Module):\n def __init__(self):\n super(Net, self).__init__()\n # convolutional layer (sees 32x32x3 image tensor)\n self.conv1 = nn.Conv2d(3, 16, 3, padding=1)\n # convolutional layer (sees 16x16x16 tensor)\n self.conv2 = nn.Conv2d(16, 32, 3, padding=1)\n # convolutional layer (sees 8x8x32 tensor)\n self.conv3 = nn.Conv2d(32, 64, 3, padding=1)\n # max pooling layer\n self.pool = nn.MaxPool2d(2, 2)\n # linear layer (64 * 4 * 4 -> 500)\n self.fc1 = nn.Linear(64 * 4 * 4, 500)\n # linear layer (500 -> 10)\n self.fc2 = nn.Linear(500, 10)\n # dropout layer (p=0.25)\n self.dropout = nn.Dropout(0.25)\n\n def forward(self, x):\n # add sequence of convolutional and max pooling layers\n x = self.pool(F.relu(self.conv1(x)))\n x = self.pool(F.relu(self.conv2(x)))\n x = self.pool(F.relu(self.conv3(x)))\n # flatten image input\n x = x.view(-1, 64 * 4 * 4)\n # add dropout layer\n x = self.dropout(x)\n # add 1st hidden layer, with relu activation function\n x = F.relu(self.fc1(x))\n # add dropout layer\n x = self.dropout(x)\n # add 2nd hidden layer, with relu activation function\n x = self.fc2(x)\n return x\n\n# create a complete CNN\nmodel = Net()\nprint(model)\n\n# move tensors to GPU if CUDA is available\nif train_on_gpu:\n model.cuda()", "Net(\n (conv1): Conv2d(3, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n (conv2): Conv2d(16, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n (conv3): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n (fc1): Linear(in_features=1024, out_features=500, bias=True)\n (fc2): Linear(in_features=500, out_features=10, bias=True)\n (dropout): Dropout(p=0.25)\n)\n" ] ], [ [ "### Specify [Loss Function](http://pytorch.org/docs/stable/nn.html#loss-functions) and [Optimizer](http://pytorch.org/docs/stable/optim.html)\n\nDecide on a loss and optimization function that is best suited for this classification task. The linked code examples from above, may be a good starting point; [this PyTorch classification example](https://github.com/pytorch/tutorials/blob/master/beginner_source/blitz/cifar10_tutorial.py) or [this, more complex Keras example](https://github.com/keras-team/keras/blob/master/examples/cifar10_cnn.py). Pay close attention to the value for **learning rate** as this value determines how your model converges to a small error.\n\n#### TODO: Define the loss and optimizer and see how these choices change the loss over time.", "_____no_output_____" ] ], [ [ "import torch.optim as optim\n\n# specify loss function (categorical cross-entropy)\ncriterion = nn.CrossEntropyLoss()\n\n# specify optimizer\noptimizer = optim.SGD(model.parameters(), lr=0.01)", "_____no_output_____" ] ], [ [ "---\n## Train the Network\n\nRemember to look at how the training and validation loss decreases over time; if the validation loss ever increases it indicates possible overfitting.", "_____no_output_____" ] ], [ [ "# number of epochs to train the model\nn_epochs = 30\n\nvalid_loss_min = np.Inf # track change in validation loss\n\nfor epoch in range(1, n_epochs+1):\n\n # keep track of training and validation loss\n train_loss = 0.0\n valid_loss = 0.0\n \n ###################\n # train the model #\n ###################\n model.train()\n for batch_idx, (data, target) in enumerate(train_loader):\n # move tensors to GPU if CUDA is available\n if train_on_gpu:\n data, target = data.cuda(), target.cuda()\n # clear the gradients of all optimized variables\n optimizer.zero_grad()\n # forward pass: compute predicted outputs by passing inputs to the model\n output = model(data)\n # calculate the batch loss\n loss = criterion(output, target)\n # backward pass: compute gradient of the loss with respect to model parameters\n loss.backward()\n # perform a single optimization step (parameter update)\n optimizer.step()\n # update training loss\n train_loss += loss.item()*data.size(0)\n \n ###################### \n # validate the model #\n ######################\n model.eval()\n for batch_idx, (data, target) in enumerate(valid_loader):\n # move tensors to GPU if CUDA is available\n if train_on_gpu:\n data, target = data.cuda(), target.cuda()\n # forward pass: compute predicted outputs by passing inputs to the model\n output = model(data)\n # calculate the batch loss\n loss = criterion(output, target)\n # update average validation loss \n valid_loss += loss.item()*data.size(0)\n \n # calculate average losses\n train_loss = train_loss/len(train_loader.dataset)\n valid_loss = valid_loss/len(valid_loader.dataset)\n \n # print training/validation statistics \n print('Epoch: {} \\tTraining Loss: {:.6f} \\tValidation Loss: {:.6f}'.format(\n epoch, train_loss, valid_loss))\n \n # save model if validation loss has decreased\n if valid_loss <= valid_loss_min:\n print('Validation loss decreased ({:.6f} --> {:.6f}). Saving model ...'.format(\n valid_loss_min,\n valid_loss))\n torch.save(model.state_dict(), 'model_augmented.pt')\n valid_loss_min = valid_loss", "Epoch: 1 \tTraining Loss: 1.483400 \tValidation Loss: 0.357015\nValidation loss decreased (inf --> 0.357015). Saving model ...\nEpoch: 2 \tTraining Loss: 1.462634 \tValidation Loss: 0.355032\nValidation loss decreased (0.357015 --> 0.355032). Saving model ...\nEpoch: 3 \tTraining Loss: 1.454201 \tValidation Loss: 0.351629\nValidation loss decreased (0.355032 --> 0.351629). Saving model ...\nEpoch: 4 \tTraining Loss: 1.446663 \tValidation Loss: 0.349965\nValidation loss decreased (0.351629 --> 0.349965). Saving model ...\nEpoch: 5 \tTraining Loss: 1.444058 \tValidation Loss: 0.348746\nValidation loss decreased (0.349965 --> 0.348746). Saving model ...\nEpoch: 6 \tTraining Loss: 1.443467 \tValidation Loss: 0.350312\nEpoch: 7 \tTraining Loss: 1.435970 \tValidation Loss: 0.347319\nValidation loss decreased (0.348746 --> 0.347319). Saving model ...\nEpoch: 8 \tTraining Loss: 1.430920 \tValidation Loss: 0.347215\nValidation loss decreased (0.347319 --> 0.347215). Saving model ...\nEpoch: 9 \tTraining Loss: 1.428707 \tValidation Loss: 0.347475\nEpoch: 10 \tTraining Loss: 1.426082 \tValidation Loss: 0.344052\nValidation loss decreased (0.347215 --> 0.344052). Saving model ...\nEpoch: 11 \tTraining Loss: 1.421921 \tValidation Loss: 0.343847\nValidation loss decreased (0.344052 --> 0.343847). Saving model ...\nEpoch: 12 \tTraining Loss: 1.419442 \tValidation Loss: 0.344311\nEpoch: 13 \tTraining Loss: 1.418585 \tValidation Loss: 0.345510\nEpoch: 14 \tTraining Loss: 1.415068 \tValidation Loss: 0.342422\nValidation loss decreased (0.343847 --> 0.342422). Saving model ...\nEpoch: 15 \tTraining Loss: 1.410706 \tValidation Loss: 0.340904\nValidation loss decreased (0.342422 --> 0.340904). Saving model ...\nEpoch: 16 \tTraining Loss: 1.415896 \tValidation Loss: 0.342239\nEpoch: 17 \tTraining Loss: 1.408537 \tValidation Loss: 0.339082\nValidation loss decreased (0.340904 --> 0.339082). Saving model ...\nEpoch: 18 \tTraining Loss: 1.406269 \tValidation Loss: 0.343943\nEpoch: 19 \tTraining Loss: 1.404933 \tValidation Loss: 0.338068\nValidation loss decreased (0.339082 --> 0.338068). Saving model ...\nEpoch: 20 \tTraining Loss: 1.404783 \tValidation Loss: 0.341946\nEpoch: 21 \tTraining Loss: 1.401647 \tValidation Loss: 0.340137\nEpoch: 22 \tTraining Loss: 1.402606 \tValidation Loss: 0.339872\nEpoch: 23 \tTraining Loss: 1.398441 \tValidation Loss: 0.337148\nValidation loss decreased (0.338068 --> 0.337148). Saving model ...\nEpoch: 24 \tTraining Loss: 1.401196 \tValidation Loss: 0.335357\nValidation loss decreased (0.337148 --> 0.335357). Saving model ...\nEpoch: 25 \tTraining Loss: 1.392405 \tValidation Loss: 0.339850\nEpoch: 26 \tTraining Loss: 1.392132 \tValidation Loss: 0.339279\nEpoch: 27 \tTraining Loss: 1.387730 \tValidation Loss: 0.331665\nValidation loss decreased (0.335357 --> 0.331665). Saving model ...\nEpoch: 28 \tTraining Loss: 1.380706 \tValidation Loss: 0.340137\nEpoch: 29 \tTraining Loss: 1.374192 \tValidation Loss: 0.332150\nEpoch: 30 \tTraining Loss: 1.372476 \tValidation Loss: 0.332638\n" ] ], [ [ "### Load the Model with the Lowest Validation Loss", "_____no_output_____" ] ], [ [ "model.load_state_dict(torch.load('model_augmented.pt'))", "_____no_output_____" ] ], [ [ "---\n## Test the Trained Network\n\nTest your trained model on previously unseen data! A \"good\" result will be a CNN that gets around 70% (or more, try your best!) accuracy on these test images.", "_____no_output_____" ] ], [ [ "# track test loss\ntest_loss = 0.0\nclass_correct = list(0. for i in range(10))\nclass_total = list(0. for i in range(10))\n\nmodel.eval()\n# iterate over test data\nfor batch_idx, (data, target) in enumerate(test_loader):\n # move tensors to GPU if CUDA is available\n if train_on_gpu:\n data, target = data.cuda(), target.cuda()\n # forward pass: compute predicted outputs by passing inputs to the model\n output = model(data)\n # calculate the batch loss\n loss = criterion(output, target)\n # update test loss \n test_loss += loss.item()*data.size(0)\n # convert output probabilities to predicted class\n _, pred = torch.max(output, 1) \n # compare predictions to true label\n correct_tensor = pred.eq(target.data.view_as(pred))\n correct = np.squeeze(correct_tensor.numpy()) if not train_on_gpu else np.squeeze(correct_tensor.cpu().numpy())\n # calculate test accuracy for each object class\n for i in range(batch_size):\n label = target.data[i]\n class_correct[label] += correct[i].item()\n class_total[label] += 1\n\n# average test loss\ntest_loss = test_loss/len(test_loader.dataset)\nprint('Test Loss: {:.6f}\\n'.format(test_loss))\n\nfor i in range(10):\n if class_total[i] > 0:\n print('Test Accuracy of %5s: %2d%% (%2d/%2d)' % (\n classes[i], 100 * class_correct[i] / class_total[i],\n np.sum(class_correct[i]), np.sum(class_total[i])))\n else:\n print('Test Accuracy of %5s: N/A (no training examples)' % (classes[i]))\n\nprint('\\nTest Accuracy (Overall): %2d%% (%2d/%2d)' % (\n 100. * np.sum(class_correct) / np.sum(class_total),\n np.sum(class_correct), np.sum(class_total)))", "Test Loss: 1.650751\n\nTest Accuracy of airplane: 37% (374/1000)\nTest Accuracy of automobile: 47% (471/1000)\nTest Accuracy of bird: 8% (80/1000)\nTest Accuracy of cat: 20% (203/1000)\nTest Accuracy of deer: 39% (390/1000)\nTest Accuracy of dog: 27% (274/1000)\nTest Accuracy of frog: 37% (372/1000)\nTest Accuracy of horse: 48% (483/1000)\nTest Accuracy of ship: 47% (473/1000)\nTest Accuracy of truck: 58% (581/1000)\n\nTest Accuracy (Overall): 37% (3701/10000)\n" ] ], [ [ "### Visualize Sample Test Results", "_____no_output_____" ] ], [ [ "# obtain one batch of test images\ndataiter = iter(test_loader)\nimages, labels = dataiter.next()\nimages.numpy()\n\n# move model inputs to cuda, if GPU available\nif train_on_gpu:\n images = images.cuda()\n\n# get sample outputs\noutput = model(images)\n# convert output probabilities to predicted class\n_, preds_tensor = torch.max(output, 1)\npreds = np.squeeze(preds_tensor.numpy()) if not train_on_gpu else np.squeeze(preds_tensor.cpu().numpy())\n\n# plot the images in the batch, along with predicted and true labels\nfig = plt.figure(figsize=(25, 4))\nfor idx in np.arange(20):\n ax = fig.add_subplot(2, 20/2, idx+1, xticks=[], yticks=[])\n imshow(images[idx])\n ax.set_title(\"{} ({})\".format(classes[preds[idx]], classes[labels[idx]]),\n color=(\"green\" if preds[idx]==labels[idx].item() else \"red\"))", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "#hide\n#skip\n! [ -e /content ] && pip install -Uqq fastai # upgrade fastai on colab", "_____no_output_____" ], [ "#default_exp data.core", "_____no_output_____" ], [ "#export\nfrom fastai.torch_basics import *\nfrom fastai.data.load import *", "_____no_output_____" ], [ "#hide\nfrom nbdev.showdoc import *", "_____no_output_____" ] ], [ [ "# Data core\n\n> Core functionality for gathering data", "_____no_output_____" ], [ "The classes here provide functionality for applying a list of transforms to a set of items (`TfmdLists`, `Datasets`) or a `DataLoader` (`TfmdDl`) as well as the base class used to gather the data for model training: `DataLoaders`.", "_____no_output_____" ], [ "## TfmdDL -", "_____no_output_____" ] ], [ [ "#export\n@typedispatch\ndef show_batch(x, y, samples, ctxs=None, max_n=9, **kwargs):\n if ctxs is None: ctxs = Inf.nones\n if hasattr(samples[0], 'show'):\n ctxs = [s.show(ctx=c, **kwargs) for s,c,_ in zip(samples,ctxs,range(max_n))]\n else:\n for i in range_of(samples[0]):\n ctxs = [b.show(ctx=c, **kwargs) for b,c,_ in zip(samples.itemgot(i),ctxs,range(max_n))]\n return ctxs", "_____no_output_____" ] ], [ [ "`show_batch` is a type-dispatched function that is responsible for showing decoded `samples`. `x` and `y` are the input and the target in the batch to be shown, and are passed along to dispatch on their types. There is a different implementation of `show_batch` if `x` is a `TensorImage` or a `TensorText` for instance (see vision.core or text.data for more details). `ctxs` can be passed but the function is responsible to create them if necessary. `kwargs` depend on the specific implementation.", "_____no_output_____" ] ], [ [ "#export\n@typedispatch\ndef show_results(x, y, samples, outs, ctxs=None, max_n=9, **kwargs):\n if ctxs is None: ctxs = Inf.nones\n for i in range(len(samples[0])):\n ctxs = [b.show(ctx=c, **kwargs) for b,c,_ in zip(samples.itemgot(i),ctxs,range(max_n))]\n for i in range(len(outs[0])):\n ctxs = [b.show(ctx=c, **kwargs) for b,c,_ in zip(outs.itemgot(i),ctxs,range(max_n))]\n return ctxs", "_____no_output_____" ] ], [ [ "`show_results` is a type-dispatched function that is responsible for showing decoded `samples` and their corresponding `outs`. Like in `show_batch`, `x` and `y` are the input and the target in the batch to be shown, and are passed along to dispatch on their types. `ctxs` can be passed but the function is responsible to create them if necessary. `kwargs` depend on the specific implementation.", "_____no_output_____" ] ], [ [ "#export\n_all_ = [\"show_batch\", \"show_results\"]", "_____no_output_____" ], [ "#export\n_batch_tfms = ('after_item','before_batch','after_batch')", "_____no_output_____" ], [ "#export\n@log_args(but_as=DataLoader.__init__)\n@delegates()\nclass TfmdDL(DataLoader):\n \"Transformed `DataLoader`\"\n def __init__(self, dataset, bs=64, shuffle=False, num_workers=None, verbose=False, do_setup=True, **kwargs):\n if num_workers is None: num_workers = min(16, defaults.cpus)\n for nm in _batch_tfms: kwargs[nm] = Pipeline(kwargs.get(nm,None))\n super().__init__(dataset, bs=bs, shuffle=shuffle, num_workers=num_workers, **kwargs)\n if do_setup:\n for nm in _batch_tfms:\n pv(f\"Setting up {nm}: {kwargs[nm]}\", verbose)\n kwargs[nm].setup(self)\n\n def _one_pass(self):\n b = self.do_batch([self.do_item(0)])\n if self.device is not None: b = to_device(b, self.device)\n its = self.after_batch(b)\n self._n_inp = 1 if not isinstance(its, (list,tuple)) or len(its)==1 else len(its)-1\n self._types = explode_types(its)\n\n def _retain_dl(self,b):\n if not getattr(self, '_types', None): self._one_pass()\n return retain_types(b, typs=self._types)\n\n @delegates(DataLoader.new)\n def new(self, dataset=None, cls=None, **kwargs):\n res = super().new(dataset, cls, do_setup=False, **kwargs)\n if not hasattr(self, '_n_inp') or not hasattr(self, '_types'):\n try:\n self._one_pass()\n res._n_inp,res._types = self._n_inp,self._types\n except: print(\"Could not do one pass in your dataloader, there is something wrong in it\")\n else: res._n_inp,res._types = self._n_inp,self._types\n return res\n\n def before_iter(self):\n super().before_iter()\n split_idx = getattr(self.dataset, 'split_idx', None)\n for nm in _batch_tfms:\n f = getattr(self,nm)\n if isinstance(f,Pipeline): f.split_idx=split_idx\n\n def decode(self, b): return self.before_batch.decode(to_cpu(self.after_batch.decode(self._retain_dl(b))))\n def decode_batch(self, b, max_n=9, full=True): return self._decode_batch(self.decode(b), max_n, full)\n\n def _decode_batch(self, b, max_n=9, full=True):\n f = self.after_item.decode\n f = compose(f, partial(getattr(self.dataset,'decode',noop), full = full))\n return L(batch_to_samples(b, max_n=max_n)).map(f)\n\n def _pre_show_batch(self, b, max_n=9):\n \"Decode `b` to be ready for `show_batch`\"\n b = self.decode(b)\n if hasattr(b, 'show'): return b,None,None\n its = self._decode_batch(b, max_n, full=False)\n if not is_listy(b): b,its = [b],L((o,) for o in its)\n return detuplify(b[:self.n_inp]),detuplify(b[self.n_inp:]),its\n\n def show_batch(self, b=None, max_n=9, ctxs=None, show=True, unique=False, **kwargs):\n if unique:\n old_get_idxs = self.get_idxs\n self.get_idxs = lambda: Inf.zeros\n if b is None: b = self.one_batch()\n if not show: return self._pre_show_batch(b, max_n=max_n)\n show_batch(*self._pre_show_batch(b, max_n=max_n), ctxs=ctxs, max_n=max_n, **kwargs)\n if unique: self.get_idxs = old_get_idxs\n\n def show_results(self, b, out, max_n=9, ctxs=None, show=True, **kwargs):\n x,y,its = self.show_batch(b, max_n=max_n, show=False)\n b_out = type(b)(b[:self.n_inp] + (tuple(out) if is_listy(out) else (out,)))\n x1,y1,outs = self.show_batch(b_out, max_n=max_n, show=False)\n res = (x,x1,None,None) if its is None else (x, y, its, outs.itemgot(slice(self.n_inp,None)))\n if not show: return res\n show_results(*res, ctxs=ctxs, max_n=max_n, **kwargs)\n\n @property\n def n_inp(self):\n if hasattr(self.dataset, 'n_inp'): return self.dataset.n_inp\n if not hasattr(self, '_n_inp'): self._one_pass()\n return self._n_inp\n\n def to(self, device):\n self.device = device\n for tfm in self.after_batch.fs:\n for a in L(getattr(tfm, 'parameters', None)): setattr(tfm, a, getattr(tfm, a).to(device))\n return self", "_____no_output_____" ] ], [ [ "A `TfmdDL` is a `DataLoader` that creates `Pipeline` from a list of `Transform`s for the callbacks `after_item`, `before_batch` and `after_batch`. As a result, it can decode or show a processed `batch`.", "_____no_output_____" ] ], [ [ "#export\nadd_docs(TfmdDL,\n decode=\"Decode `b` using `tfms`\",\n decode_batch=\"Decode `b` entirely\",\n new=\"Create a new version of self with a few changed attributes\",\n show_batch=\"Show `b` (defaults to `one_batch`), a list of lists of pipeline outputs (i.e. output of a `DataLoader`)\",\n show_results=\"Show each item of `b` and `out`\",\n before_iter=\"override\",\n to=\"Put self and its transforms state on `device`\")", "_____no_output_____" ], [ "class _Category(int, ShowTitle): pass", "_____no_output_____" ], [ "#Test retain type\nclass NegTfm(Transform):\n def encodes(self, x): return torch.neg(x)\n def decodes(self, x): return torch.neg(x)\n \ntdl = TfmdDL([(TensorImage([1]),)] * 4, after_batch=NegTfm(), bs=4, num_workers=4)\nb = tdl.one_batch()\ntest_eq(type(b[0]), TensorImage)\nb = (tensor([1.,1.,1.,1.]),)\ntest_eq(type(tdl.decode_batch(b)[0][0]), TensorImage)", "_____no_output_____" ], [ "class A(Transform): \n def encodes(self, x): return x \n def decodes(self, x): return TitledInt(x) \n\n@Transform\ndef f(x)->None: return fastuple((x,x))\n\nstart = torch.arange(50)\ntest_eq_type(f(2), fastuple((2,2)))", "_____no_output_____" ], [ "a = A()\ntdl = TfmdDL(start, after_item=lambda x: (a(x), f(x)), bs=4)\nx,y = tdl.one_batch()\ntest_eq(type(y), fastuple)\n\ns = tdl.decode_batch((x,y))\ntest_eq(type(s[0][1]), fastuple)", "_____no_output_____" ], [ "tdl = TfmdDL(torch.arange(0,50), after_item=A(), after_batch=NegTfm(), bs=4)\ntest_eq(tdl.dataset[0], start[0])\ntest_eq(len(tdl), (50-1)//4+1)\ntest_eq(tdl.bs, 4)\ntest_stdout(tdl.show_batch, '0\\n1\\n2\\n3')\ntest_stdout(partial(tdl.show_batch, unique=True), '0\\n0\\n0\\n0')", "_____no_output_____" ], [ "class B(Transform):\n parameters = 'a'\n def __init__(self): self.a = torch.tensor(0.)\n def encodes(self, x): x\n \ntdl = TfmdDL([(TensorImage([1]),)] * 4, after_batch=B(), bs=4)\ntest_eq(tdl.after_batch.fs[0].a.device, torch.device('cpu'))\ntdl.to(default_device())\ntest_eq(tdl.after_batch.fs[0].a.device, default_device())", "_____no_output_____" ] ], [ [ "### Methods", "_____no_output_____" ] ], [ [ "show_doc(TfmdDL.one_batch)", "_____no_output_____" ], [ "tfm = NegTfm()\ntdl = TfmdDL(start, after_batch=tfm, bs=4)", "_____no_output_____" ], [ "b = tdl.one_batch()\ntest_eq(tensor([0,-1,-2,-3]), b)", "_____no_output_____" ], [ "show_doc(TfmdDL.decode)", "_____no_output_____" ], [ "test_eq(tdl.decode(b), tensor(0,1,2,3))", "_____no_output_____" ], [ "show_doc(TfmdDL.decode_batch)", "_____no_output_____" ], [ "test_eq(tdl.decode_batch(b), [0,1,2,3])", "_____no_output_____" ], [ "show_doc(TfmdDL.show_batch)", "_____no_output_____" ], [ "show_doc(TfmdDL.to)", "_____no_output_____" ] ], [ [ "## DataLoaders -", "_____no_output_____" ] ], [ [ "# export\n@docs\nclass DataLoaders(GetAttr):\n \"Basic wrapper around several `DataLoader`s.\"\n _default='train'\n def __init__(self, *loaders, path='.', device=None):\n self.loaders,self.path = list(loaders),Path(path)\n if device is not None or hasattr(loaders[0],'to'): self.device = device\n\n def __getitem__(self, i): return self.loaders[i]\n def new_empty(self):\n loaders = [dl.new(dl.dataset.new_empty()) for dl in self.loaders]\n return type(self)(*loaders, path=self.path, device=self.device)\n\n def _set(i, self, v): self.loaders[i] = v\n train ,valid = add_props(lambda i,x: x[i], _set)\n train_ds,valid_ds = add_props(lambda i,x: x[i].dataset)\n\n @property\n def device(self): return self._device\n\n @device.setter\n def device(self, d):\n for dl in self.loaders: dl.to(d)\n self._device = d\n\n def to(self, device):\n self.device = device\n return self\n\n def cuda(self): return self.to(device=default_device())\n def cpu(self): return self.to(device=torch.device('cpu'))\n\n @classmethod\n def from_dsets(cls, *ds, path='.', bs=64, device=None, dl_type=TfmdDL, **kwargs):\n default = (True,) + (False,) * (len(ds)-1)\n defaults = {'shuffle': default, 'drop_last': default}\n for nm in _batch_tfms:\n if nm in kwargs: kwargs[nm] = Pipeline(kwargs[nm])\n kwargs = merge(defaults, {k: tuplify(v, match=ds) for k,v in kwargs.items()})\n kwargs = [{k: v[i] for k,v in kwargs.items()} for i in range_of(ds)]\n return cls(*[dl_type(d, bs=bs, **k) for d,k in zip(ds, kwargs)], path=path, device=device)\n\n @classmethod\n def from_dblock(cls, dblock, source, path='.', bs=64, val_bs=None, shuffle_train=True, device=None, **kwargs):\n return dblock.dataloaders(source, path=path, bs=bs, val_bs=val_bs, shuffle_train=shuffle_train, device=device, **kwargs)\n\n _docs=dict(__getitem__=\"Retrieve `DataLoader` at `i` (`0` is training, `1` is validation)\",\n train=\"Training `DataLoader`\",\n valid=\"Validation `DataLoader`\",\n train_ds=\"Training `Dataset`\",\n valid_ds=\"Validation `Dataset`\",\n to=\"Use `device`\",\n cuda=\"Use the gpu if available\",\n cpu=\"Use the cpu\",\n new_empty=\"Create a new empty version of `self` with the same transforms\",\n from_dblock=\"Create a dataloaders from a given `dblock`\")", "_____no_output_____" ], [ "dls = DataLoaders(tdl,tdl)\nx = dls.train.one_batch()\nx2 = first(tdl)\ntest_eq(x,x2)\nx2 = dls.one_batch()\ntest_eq(x,x2)", "_____no_output_____" ], [ "#hide\n#test assignment works\ndls.train = dls.train.new(bs=4)", "_____no_output_____" ] ], [ [ "### Methods", "_____no_output_____" ] ], [ [ "show_doc(DataLoaders.__getitem__)", "_____no_output_____" ], [ "x2 = dls[0].one_batch()\ntest_eq(x,x2)", "_____no_output_____" ], [ "show_doc(DataLoaders.train, name=\"DataLoaders.train\")", "_____no_output_____" ], [ "show_doc(DataLoaders.valid, name=\"DataLoaders.valid\")", "_____no_output_____" ], [ "show_doc(DataLoaders.train_ds, name=\"DataLoaders.train_ds\")", "_____no_output_____" ], [ "show_doc(DataLoaders.valid_ds, name=\"DataLoaders.valid_ds\")", "_____no_output_____" ] ], [ [ "## TfmdLists -", "_____no_output_____" ] ], [ [ "#export\nclass FilteredBase:\n \"Base class for lists with subsets\"\n _dl_type,_dbunch_type = TfmdDL,DataLoaders\n def __init__(self, *args, dl_type=None, **kwargs):\n if dl_type is not None: self._dl_type = dl_type\n self.dataloaders = delegates(self._dl_type.__init__)(self.dataloaders)\n super().__init__(*args, **kwargs)\n\n @property\n def n_subsets(self): return len(self.splits)\n def _new(self, items, **kwargs): return super()._new(items, splits=self.splits, **kwargs)\n def subset(self): raise NotImplemented\n\n def dataloaders(self, bs=64, val_bs=None, shuffle_train=True, n=None, path='.', dl_type=None, dl_kwargs=None,\n device=None, **kwargs):\n if device is None: device=default_device()\n if dl_kwargs is None: dl_kwargs = [{}] * self.n_subsets\n if dl_type is None: dl_type = self._dl_type\n drop_last = kwargs.pop('drop_last', shuffle_train)\n dl = dl_type(self.subset(0), bs=bs, shuffle=shuffle_train, drop_last=drop_last, n=n, device=device,\n **merge(kwargs, dl_kwargs[0]))\n dls = [dl] + [dl.new(self.subset(i), bs=(bs if val_bs is None else val_bs), shuffle=False, drop_last=False,\n n=None, **dl_kwargs[i]) for i in range(1, self.n_subsets)]\n return self._dbunch_type(*dls, path=path, device=device)\n\nFilteredBase.train,FilteredBase.valid = add_props(lambda i,x: x.subset(i))", "_____no_output_____" ], [ "#export\nclass TfmdLists(FilteredBase, L, GetAttr):\n \"A `Pipeline` of `tfms` applied to a collection of `items`\"\n _default='tfms'\n def __init__(self, items, tfms, use_list=None, do_setup=True, split_idx=None, train_setup=True,\n splits=None, types=None, verbose=False, dl_type=None):\n super().__init__(items, use_list=use_list)\n if dl_type is not None: self._dl_type = dl_type\n self.splits = L([slice(None),[]] if splits is None else splits).map(mask2idxs)\n if isinstance(tfms,TfmdLists): tfms = tfms.tfms\n if isinstance(tfms,Pipeline): do_setup=False\n self.tfms = Pipeline(tfms, split_idx=split_idx)\n store_attr('types,split_idx')\n if do_setup:\n pv(f\"Setting up {self.tfms}\", verbose)\n self.setup(train_setup=train_setup)\n\n def _new(self, items, split_idx=None, **kwargs):\n split_idx = ifnone(split_idx,self.split_idx)\n return super()._new(items, tfms=self.tfms, do_setup=False, types=self.types, split_idx=split_idx, **kwargs)\n def subset(self, i): return self._new(self._get(self.splits[i]), split_idx=i)\n def _after_item(self, o): return self.tfms(o)\n def __repr__(self): return f\"{self.__class__.__name__}: {self.items}\\ntfms - {self.tfms.fs}\"\n def __iter__(self): return (self[i] for i in range(len(self)))\n def show(self, o, **kwargs): return self.tfms.show(o, **kwargs)\n def decode(self, o, **kwargs): return self.tfms.decode(o, **kwargs)\n def __call__(self, o, **kwargs): return self.tfms.__call__(o, **kwargs)\n def overlapping_splits(self): return L(Counter(self.splits.concat()).values()).filter(gt(1))\n def new_empty(self): return self._new([])\n\n def setup(self, train_setup=True):\n self.tfms.setup(self, train_setup)\n if len(self) != 0:\n x = super().__getitem__(0) if self.splits is None else super().__getitem__(self.splits[0])[0]\n self.types = []\n for f in self.tfms.fs:\n self.types.append(getattr(f, 'input_types', type(x)))\n x = f(x)\n self.types.append(type(x))\n types = L(t if is_listy(t) else [t] for t in self.types).concat().unique()\n self.pretty_types = '\\n'.join([f' - {t}' for t in types])\n\n def infer_idx(self, x):\n # TODO: check if we really need this, or can simplify\n idx = 0\n for t in self.types:\n if isinstance(x, t): break\n idx += 1\n types = L(t if is_listy(t) else [t] for t in self.types).concat().unique()\n pretty_types = '\\n'.join([f' - {t}' for t in types])\n assert idx < len(self.types), f\"Expected an input of type in \\n{pretty_types}\\n but got {type(x)}\"\n return idx\n\n def infer(self, x):\n return compose_tfms(x, tfms=self.tfms.fs[self.infer_idx(x):], split_idx=self.split_idx)\n\n def __getitem__(self, idx):\n res = super().__getitem__(idx)\n if self._after_item is None: return res\n return self._after_item(res) if is_indexer(idx) else res.map(self._after_item)", "_____no_output_____" ], [ "#export\nadd_docs(TfmdLists,\n setup=\"Transform setup with self\",\n decode=\"From `Pipeline`\",\n show=\"From `Pipeline`\",\n overlapping_splits=\"All splits that are in more than one split\",\n subset=\"New `TfmdLists` with same tfms that only includes items in `i`th split\",\n infer_idx=\"Finds the index where `self.tfms` can be applied to `x`, depending on the type of `x`\",\n infer=\"Apply `self.tfms` to `x` starting at the right tfm depending on the type of `x`\",\n new_empty=\"A new version of `self` but with no items\")", "_____no_output_____" ], [ "#exports\ndef decode_at(o, idx):\n \"Decoded item at `idx`\"\n return o.decode(o[idx])", "_____no_output_____" ], [ "#exports\ndef show_at(o, idx, **kwargs):\n \"Show item at `idx`\",\n return o.show(o[idx], **kwargs)", "_____no_output_____" ] ], [ [ "A `TfmdLists` combines a collection of object with a `Pipeline`. `tfms` can either be a `Pipeline` or a list of transforms, in which case, it will wrap them in a `Pipeline`. `use_list` is passed along to `L` with the `items` and `split_idx` are passed to each transform of the `Pipeline`. `do_setup` indicates if the `Pipeline.setup` method should be called during initialization.", "_____no_output_____" ] ], [ [ "class _IntFloatTfm(Transform):\n def encodes(self, o): return TitledInt(o)\n def decodes(self, o): return TitledFloat(o)\nint2f_tfm=_IntFloatTfm()\n\ndef _neg(o): return -o\nneg_tfm = Transform(_neg, _neg)", "_____no_output_____" ], [ "items = L([1.,2.,3.]); tfms = [neg_tfm, int2f_tfm]\ntl = TfmdLists(items, tfms=tfms)\ntest_eq_type(tl[0], TitledInt(-1))\ntest_eq_type(tl[1], TitledInt(-2))\ntest_eq_type(tl.decode(tl[2]), TitledFloat(3.))\ntest_stdout(lambda: show_at(tl, 2), '-3')\ntest_eq(tl.types, [float, float, TitledInt])\ntl", "_____no_output_____" ], [ "# add splits to TfmdLists\nsplits = [[0,2],[1]]\ntl = TfmdLists(items, tfms=tfms, splits=splits)\ntest_eq(tl.n_subsets, 2)\ntest_eq(tl.train, tl.subset(0))\ntest_eq(tl.valid, tl.subset(1))\ntest_eq(tl.train.items, items[splits[0]])\ntest_eq(tl.valid.items, items[splits[1]])\ntest_eq(tl.train.tfms.split_idx, 0)\ntest_eq(tl.valid.tfms.split_idx, 1)\ntest_eq(tl.train.new_empty().split_idx, 0)\ntest_eq(tl.valid.new_empty().split_idx, 1)\ntest_eq_type(tl.splits, L(splits))\nassert not tl.overlapping_splits()", "_____no_output_____" ], [ "df = pd.DataFrame(dict(a=[1,2,3],b=[2,3,4]))\ntl = TfmdLists(df, lambda o: o.a+1, splits=[[0],[1,2]])\ntest_eq(tl[1,2], [3,4])\ntr = tl.subset(0)\ntest_eq(tr[:], [2])\nval = tl.subset(1)\ntest_eq(val[:], [3,4])", "_____no_output_____" ], [ "class _B(Transform):\n def __init__(self): self.m = 0\n def encodes(self, o): return o+self.m\n def decodes(self, o): return o-self.m\n def setups(self, items): \n print(items)\n self.m = tensor(items).float().mean().item()\n\n# test for setup, which updates `self.m`\ntl = TfmdLists(items, _B())\ntest_eq(tl.m, 2)", "TfmdLists: [1.0, 2.0, 3.0]\ntfms - (#0) []\n" ] ], [ [ "Here's how we can use `TfmdLists.setup` to implement a simple category list, getting labels from a mock file list:", "_____no_output_____" ] ], [ [ "class _Cat(Transform):\n order = 1\n def encodes(self, o): return int(self.o2i[o])\n def decodes(self, o): return TitledStr(self.vocab[o])\n def setups(self, items): self.vocab,self.o2i = uniqueify(L(items), sort=True, bidir=True)\ntcat = _Cat()\n\ndef _lbl(o): return TitledStr(o.split('_')[0])\n\n# Check that tfms are sorted by `order` & `_lbl` is called first\nfns = ['dog_0.jpg','cat_0.jpg','cat_2.jpg','cat_1.jpg','dog_1.jpg']\ntl = TfmdLists(fns, [tcat,_lbl])\nexp_voc = ['cat','dog']\ntest_eq(tcat.vocab, exp_voc)\ntest_eq(tl.tfms.vocab, exp_voc)\ntest_eq(tl.vocab, exp_voc)\ntest_eq(tl, (1,0,0,0,1))\ntest_eq([tl.decode(o) for o in tl], ('dog','cat','cat','cat','dog'))", "_____no_output_____" ], [ "#Check only the training set is taken into account for setup\ntl = TfmdLists(fns, [tcat,_lbl], splits=[[0,4], [1,2,3]])\ntest_eq(tcat.vocab, ['dog'])", "_____no_output_____" ], [ "tfm = NegTfm(split_idx=1)\ntds = TfmdLists(start, A())\ntdl = TfmdDL(tds, after_batch=tfm, bs=4)\nx = tdl.one_batch()\ntest_eq(x, torch.arange(4))\ntds.split_idx = 1\nx = tdl.one_batch()\ntest_eq(x, -torch.arange(4))\ntds.split_idx = 0\nx = tdl.one_batch()\ntest_eq(x, torch.arange(4))", "_____no_output_____" ], [ "tds = TfmdLists(start, A())\ntdl = TfmdDL(tds, after_batch=NegTfm(), bs=4)\ntest_eq(tdl.dataset[0], start[0])\ntest_eq(len(tdl), (len(tds)-1)//4+1)\ntest_eq(tdl.bs, 4)\ntest_stdout(tdl.show_batch, '0\\n1\\n2\\n3')", "_____no_output_____" ], [ "show_doc(TfmdLists.subset)", "_____no_output_____" ], [ "show_doc(TfmdLists.infer_idx)", "_____no_output_____" ], [ "show_doc(TfmdLists.infer)", "_____no_output_____" ], [ "def mult(x): return x*2\nmult.order = 2\n\nfns = ['dog_0.jpg','cat_0.jpg','cat_2.jpg','cat_1.jpg','dog_1.jpg']\ntl = TfmdLists(fns, [_lbl,_Cat(),mult])\n\ntest_eq(tl.infer_idx('dog_45.jpg'), 0)\ntest_eq(tl.infer('dog_45.jpg'), 2)\n\ntest_eq(tl.infer_idx(4), 2)\ntest_eq(tl.infer(4), 8)\n\ntest_fail(lambda: tl.infer_idx(2.0))\ntest_fail(lambda: tl.infer(2.0))", "_____no_output_____" ], [ "#hide\n#Test input_types works on a Transform\ncat = _Cat()\ncat.input_types = (str, float)\ntl = TfmdLists(fns, [_lbl,cat,mult])\ntest_eq(tl.infer_idx(2.0), 1)\n\n#Test type annotations work on a function\ndef mult(x:(int,float)): return x*2\nmult.order = 2\ntl = TfmdLists(fns, [_lbl,_Cat(),mult])\ntest_eq(tl.infer_idx(2.0), 2)", "_____no_output_____" ] ], [ [ "## Datasets -", "_____no_output_____" ] ], [ [ "#export\n@docs\n@delegates(TfmdLists)\nclass Datasets(FilteredBase):\n \"A dataset that creates a tuple from each `tfms`, passed through `item_tfms`\"\n def __init__(self, items=None, tfms=None, tls=None, n_inp=None, dl_type=None, **kwargs):\n super().__init__(dl_type=dl_type)\n self.tls = L(tls if tls else [TfmdLists(items, t, **kwargs) for t in L(ifnone(tfms,[None]))])\n self.n_inp = ifnone(n_inp, max(1, len(self.tls)-1))\n\n def __getitem__(self, it):\n res = tuple([tl[it] for tl in self.tls])\n return res if is_indexer(it) else list(zip(*res))\n\n def __getattr__(self,k): return gather_attrs(self, k, 'tls')\n def __dir__(self): return super().__dir__() + gather_attr_names(self, 'tls')\n def __len__(self): return len(self.tls[0])\n def __iter__(self): return (self[i] for i in range(len(self)))\n def __repr__(self): return coll_repr(self)\n def decode(self, o, full=True): return tuple(tl.decode(o_, full=full) for o_,tl in zip(o,tuplify(self.tls, match=o)))\n def subset(self, i): return type(self)(tls=L(tl.subset(i) for tl in self.tls), n_inp=self.n_inp)\n def _new(self, items, *args, **kwargs): return super()._new(items, tfms=self.tfms, do_setup=False, **kwargs)\n def overlapping_splits(self): return self.tls[0].overlapping_splits()\n def new_empty(self): return type(self)(tls=[tl.new_empty() for tl in self.tls], n_inp=self.n_inp)\n @property\n def splits(self): return self.tls[0].splits\n @property\n def split_idx(self): return self.tls[0].tfms.split_idx\n @property\n def items(self): return self.tls[0].items\n @items.setter\n def items(self, v):\n for tl in self.tls: tl.items = v\n\n def show(self, o, ctx=None, **kwargs):\n for o_,tl in zip(o,self.tls): ctx = tl.show(o_, ctx=ctx, **kwargs)\n return ctx\n\n @contextmanager\n def set_split_idx(self, i):\n old_split_idx = self.split_idx\n for tl in self.tls: tl.tfms.split_idx = i\n try: yield self\n finally:\n for tl in self.tls: tl.tfms.split_idx = old_split_idx\n\n _docs=dict(\n decode=\"Compose `decode` of all `tuple_tfms` then all `tfms` on `i`\",\n show=\"Show item `o` in `ctx`\",\n dataloaders=\"Get a `DataLoaders`\",\n overlapping_splits=\"All splits that are in more than one split\",\n subset=\"New `Datasets` that only includes subset `i`\",\n new_empty=\"Create a new empty version of the `self`, keeping only the transforms\",\n set_split_idx=\"Contextmanager to use the same `Datasets` with another `split_idx`\"\n )", "_____no_output_____" ] ], [ [ "A `Datasets` creates a tuple from `items` (typically input,target) by applying to them each list of `Transform` (or `Pipeline`) in `tfms`. Note that if `tfms` contains only one list of `tfms`, the items given by `Datasets` will be tuples of one element. \n\n`n_inp` is the number of elements in the tuples that should be considered part of the input and will default to 1 if `tfms` consists of one set of transforms, `len(tfms)-1` otherwise. In most cases, the number of elements in the tuples spit out by `Datasets` will be 2 (for input,target) but it can happen that there is 3 (Siamese networks or tabular data) in which case we need to be able to determine when the inputs end and the targets begin.", "_____no_output_____" ] ], [ [ "items = [1,2,3,4]\ndsets = Datasets(items, [[neg_tfm,int2f_tfm], [add(1)]])\nt = dsets[0]\ntest_eq(t, (-1,2))\ntest_eq(dsets[0,1,2], [(-1,2),(-2,3),(-3,4)])\ntest_eq(dsets.n_inp, 1)\ndsets.decode(t)", "_____no_output_____" ], [ "class Norm(Transform):\n def encodes(self, o): return (o-self.m)/self.s\n def decodes(self, o): return (o*self.s)+self.m\n def setups(self, items):\n its = tensor(items).float()\n self.m,self.s = its.mean(),its.std()", "_____no_output_____" ], [ "items = [1,2,3,4]\nnrm = Norm()\ndsets = Datasets(items, [[neg_tfm,int2f_tfm], [neg_tfm,nrm]])\n\nx,y = zip(*dsets)\ntest_close(tensor(y).mean(), 0)\ntest_close(tensor(y).std(), 1)\ntest_eq(x, (-1,-2,-3,-4,))\ntest_eq(nrm.m, -2.5)\ntest_stdout(lambda:show_at(dsets, 1), '-2')\n\ntest_eq(dsets.m, nrm.m)\ntest_eq(dsets.norm.m, nrm.m)\ntest_eq(dsets.train.norm.m, nrm.m)", "_____no_output_____" ], [ "#hide\n#Check filtering is properly applied\nclass B(Transform):\n def encodes(self, x)->None: return int(x+1)\n def decodes(self, x): return TitledInt(x-1)\nadd1 = B(split_idx=1)\n\ndsets = Datasets(items, [neg_tfm, [neg_tfm,int2f_tfm,add1]], splits=[[3],[0,1,2]])\ntest_eq(dsets[1], [-2,-2])\ntest_eq(dsets.valid[1], [-2,-1])\ntest_eq(dsets.valid[[1,1]], [[-2,-1], [-2,-1]])\ntest_eq(dsets.train[0], [-4,-4])", "_____no_output_____" ], [ "test_fns = ['dog_0.jpg','cat_0.jpg','cat_2.jpg','cat_1.jpg','kid_1.jpg']\ntcat = _Cat()\ndsets = Datasets(test_fns, [[tcat,_lbl]], splits=[[0,1,2], [3,4]])\ntest_eq(tcat.vocab, ['cat','dog'])\ntest_eq(dsets.train, [(1,),(0,),(0,)])\ntest_eq(dsets.valid[0], (0,))\ntest_stdout(lambda: show_at(dsets.train, 0), \"dog\")", "_____no_output_____" ], [ "inp = [0,1,2,3,4]\ndsets = Datasets(inp, tfms=[None])\n\ntest_eq(*dsets[2], 2) # Retrieve one item (subset 0 is the default)\ntest_eq(dsets[1,2], [(1,),(2,)]) # Retrieve two items by index\nmask = [True,False,False,True,False]\ntest_eq(dsets[mask], [(0,),(3,)]) # Retrieve two items by mask", "_____no_output_____" ], [ "inp = pd.DataFrame(dict(a=[5,1,2,3,4]))\ndsets = Datasets(inp, tfms=attrgetter('a')).subset(0)\ntest_eq(*dsets[2], 2) # Retrieve one item (subset 0 is the default)\ntest_eq(dsets[1,2], [(1,),(2,)]) # Retrieve two items by index\nmask = [True,False,False,True,False]\ntest_eq(dsets[mask], [(5,),(3,)]) # Retrieve two items by mask", "_____no_output_____" ], [ "#test n_inp\ninp = [0,1,2,3,4]\ndsets = Datasets(inp, tfms=[None])\ntest_eq(dsets.n_inp, 1)\ndsets = Datasets(inp, tfms=[[None],[None],[None]])\ntest_eq(dsets.n_inp, 2)\ndsets = Datasets(inp, tfms=[[None],[None],[None]], n_inp=1)\ntest_eq(dsets.n_inp, 1)", "_____no_output_____" ], [ "# splits can be indices\ndsets = Datasets(range(5), tfms=[None], splits=[tensor([0,2]), [1,3,4]])\n\ntest_eq(dsets.subset(0), [(0,),(2,)])\ntest_eq(dsets.train, [(0,),(2,)]) # Subset 0 is aliased to `train`\ntest_eq(dsets.subset(1), [(1,),(3,),(4,)])\ntest_eq(dsets.valid, [(1,),(3,),(4,)]) # Subset 1 is aliased to `valid`\ntest_eq(*dsets.valid[2], 4)\n#assert '[(1,),(3,),(4,)]' in str(dsets) and '[(0,),(2,)]' in str(dsets)\ndsets", "_____no_output_____" ], [ "# splits can be boolean masks (they don't have to cover all items, but must be disjoint)\nsplits = [[False,True,True,False,True], [True,False,False,False,False]]\ndsets = Datasets(range(5), tfms=[None], splits=splits)\n\ntest_eq(dsets.train, [(1,),(2,),(4,)])\ntest_eq(dsets.valid, [(0,)])", "_____no_output_____" ], [ "# apply transforms to all items\ntfm = [[lambda x: x*2,lambda x: x+1]]\nsplits = [[1,2],[0,3,4]]\ndsets = Datasets(range(5), tfm, splits=splits)\ntest_eq(dsets.train,[(3,),(5,)])\ntest_eq(dsets.valid,[(1,),(7,),(9,)])\ntest_eq(dsets.train[False,True], [(5,)])", "_____no_output_____" ], [ "# only transform subset 1\nclass _Tfm(Transform):\n split_idx=1\n def encodes(self, x): return x*2\n def decodes(self, x): return TitledStr(x//2)", "_____no_output_____" ], [ "dsets = Datasets(range(5), [_Tfm()], splits=[[1,2],[0,3,4]])\ntest_eq(dsets.train,[(1,),(2,)])\ntest_eq(dsets.valid,[(0,),(6,),(8,)])\ntest_eq(dsets.train[False,True], [(2,)])\ndsets", "_____no_output_____" ], [ "#A context manager to change the split_idx and apply the validation transform on the training set\nds = dsets.train\nwith ds.set_split_idx(1):\n test_eq(ds,[(2,),(4,)])\ntest_eq(dsets.train,[(1,),(2,)])", "_____no_output_____" ], [ "#hide\n#Test Datasets pickles\ndsrc1 = pickle.loads(pickle.dumps(dsets))\ntest_eq(dsets.train, dsrc1.train)\ntest_eq(dsets.valid, dsrc1.valid)", "_____no_output_____" ], [ "dsets = Datasets(range(5), [_Tfm(),noop], splits=[[1,2],[0,3,4]])\ntest_eq(dsets.train,[(1,1),(2,2)])\ntest_eq(dsets.valid,[(0,0),(6,3),(8,4)])", "_____no_output_____" ], [ "start = torch.arange(0,50)\ntds = Datasets(start, [A()])\ntdl = TfmdDL(tds, after_item=NegTfm(), bs=4)\nb = tdl.one_batch()\ntest_eq(tdl.decode_batch(b), ((0,),(1,),(2,),(3,)))\ntest_stdout(tdl.show_batch, \"0\\n1\\n2\\n3\")", "_____no_output_____" ], [ "# only transform subset 1\nclass _Tfm(Transform):\n split_idx=1\n def encodes(self, x): return x*2\n\ndsets = Datasets(range(8), [None], splits=[[1,2,5,7],[0,3,4,6]])", "_____no_output_____" ], [ "# only transform subset 1\nclass _Tfm(Transform):\n split_idx=1\n def encodes(self, x): return x*2\n\ndsets = Datasets(range(8), [None], splits=[[1,2,5,7],[0,3,4,6]])\ndls = dsets.dataloaders(bs=4, after_batch=_Tfm(), shuffle_train=False, device=torch.device('cpu'))\ntest_eq(dls.train, [(tensor([1,2,5, 7]),)])\ntest_eq(dls.valid, [(tensor([0,6,8,12]),)])\ntest_eq(dls.n_inp, 1)", "_____no_output_____" ] ], [ [ "### Methods", "_____no_output_____" ] ], [ [ "items = [1,2,3,4]\ndsets = Datasets(items, [[neg_tfm,int2f_tfm]])", "_____no_output_____" ], [ "#hide_input\n_dsrc = Datasets([1,2])\nshow_doc(_dsrc.dataloaders, name=\"Datasets.dataloaders\")", "_____no_output_____" ], [ "show_doc(Datasets.decode)", "_____no_output_____" ], [ "test_eq(*dsets[0], -1)\ntest_eq(*dsets.decode((-1,)), 1)", "_____no_output_____" ], [ "show_doc(Datasets.show)", "_____no_output_____" ], [ "test_stdout(lambda:dsets.show(dsets[1]), '-2')", "_____no_output_____" ], [ "show_doc(Datasets.new_empty)", "_____no_output_____" ], [ "items = [1,2,3,4]\nnrm = Norm()\ndsets = Datasets(items, [[neg_tfm,int2f_tfm], [neg_tfm]])\nempty = dsets.new_empty()\ntest_eq(empty.items, [])", "_____no_output_____" ], [ "#hide\n#test it works for dataframes too\ndf = pd.DataFrame({'a':[1,2,3,4,5], 'b':[6,7,8,9,10]})\ndsets = Datasets(df, [[attrgetter('a')], [attrgetter('b')]])\nempty = dsets.new_empty()", "_____no_output_____" ] ], [ [ "## Add test set for inference", "_____no_output_____" ] ], [ [ "# only transform subset 1\nclass _Tfm1(Transform):\n split_idx=0\n def encodes(self, x): return x*3\n\ndsets = Datasets(range(8), [[_Tfm(),_Tfm1()]], splits=[[1,2,5,7],[0,3,4,6]])\ntest_eq(dsets.train, [(3,),(6,),(15,),(21,)])\ntest_eq(dsets.valid, [(0,),(6,),(8,),(12,)])", "_____no_output_____" ], [ "#export\ndef test_set(dsets, test_items, rm_tfms=None, with_labels=False):\n \"Create a test set from `test_items` using validation transforms of `dsets`\"\n if isinstance(dsets, Datasets):\n tls = dsets.tls if with_labels else dsets.tls[:dsets.n_inp]\n test_tls = [tl._new(test_items, split_idx=1) for tl in tls]\n if rm_tfms is None: rm_tfms = [tl.infer_idx(get_first(test_items)) for tl in test_tls]\n else: rm_tfms = tuplify(rm_tfms, match=test_tls)\n for i,j in enumerate(rm_tfms): test_tls[i].tfms.fs = test_tls[i].tfms.fs[j:]\n return Datasets(tls=test_tls)\n elif isinstance(dsets, TfmdLists):\n test_tl = dsets._new(test_items, split_idx=1)\n if rm_tfms is None: rm_tfms = dsets.infer_idx(get_first(test_items))\n test_tl.tfms.fs = test_tl.tfms.fs[rm_tfms:]\n return test_tl\n else: raise Exception(f\"This method requires using the fastai library to assemble your data. Expected a `Datasets` or a `TfmdLists` but got {dsets.__class__.__name__}\")", "_____no_output_____" ], [ "class _Tfm1(Transform):\n split_idx=0\n def encodes(self, x): return x*3\n\ndsets = Datasets(range(8), [[_Tfm(),_Tfm1()]], splits=[[1,2,5,7],[0,3,4,6]])\ntest_eq(dsets.train, [(3,),(6,),(15,),(21,)])\ntest_eq(dsets.valid, [(0,),(6,),(8,),(12,)])\n\n#Tranform of the validation set are applied\ntst = test_set(dsets, [1,2,3])\ntest_eq(tst, [(2,),(4,),(6,)])", "_____no_output_____" ], [ "#hide\n#Test with different types\ntfm = _Tfm1()\ntfm.split_idx,tfm.order = None,2\ndsets = Datasets(['dog', 'cat', 'cat', 'dog'], [[_Cat(),tfm]])\n\n#With strings\ntest_eq(test_set(dsets, ['dog', 'cat', 'cat']), [(3,), (0,), (0,)])\n#With ints\ntest_eq(test_set(dsets, [1,2]), [(3,), (6,)])", "_____no_output_____" ], [ "#hide\n#Test with various input lengths\ndsets = Datasets(range(8), [[_Tfm(),_Tfm1()],[_Tfm(),_Tfm1()],[_Tfm(),_Tfm1()]], splits=[[1,2,5,7],[0,3,4,6]])\ntst = test_set(dsets, [1,2,3])\ntest_eq(tst, [(2,2),(4,4),(6,6)])\n\ndsets = Datasets(range(8), [[_Tfm(),_Tfm1()],[_Tfm(),_Tfm1()],[_Tfm(),_Tfm1()]], splits=[[1,2,5,7],[0,3,4,6]], n_inp=1)\ntst = test_set(dsets, [1,2,3])\ntest_eq(tst, [(2,),(4,),(6,)])", "_____no_output_____" ], [ "#hide\n#Test with rm_tfms\ndsets = Datasets(range(8), [[_Tfm(),_Tfm()]], splits=[[1,2,5,7],[0,3,4,6]])\ntst = test_set(dsets, [1,2,3])\ntest_eq(tst, [(4,),(8,),(12,)])\n\ndsets = Datasets(range(8), [[_Tfm(),_Tfm()]], splits=[[1,2,5,7],[0,3,4,6]])\ntst = test_set(dsets, [1,2,3], rm_tfms=1)\ntest_eq(tst, [(2,),(4,),(6,)])\n\ndsets = Datasets(range(8), [[_Tfm(),_Tfm()], [_Tfm(),_Tfm()]], splits=[[1,2,5,7],[0,3,4,6]], n_inp=2)\ntst = test_set(dsets, [1,2,3], rm_tfms=(1,0))\ntest_eq(tst, [(2,4),(4,8),(6,12)])", "_____no_output_____" ], [ "#export\n@delegates(TfmdDL.__init__)\n@patch\ndef test_dl(self:DataLoaders, test_items, rm_type_tfms=None, with_labels=False, **kwargs):\n \"Create a test dataloader from `test_items` using validation transforms of `dls`\"\n test_ds = test_set(self.valid_ds, test_items, rm_tfms=rm_type_tfms, with_labels=with_labels\n ) if isinstance(self.valid_ds, (Datasets, TfmdLists)) else test_items\n return self.valid.new(test_ds, **kwargs)", "_____no_output_____" ], [ "dsets = Datasets(range(8), [[_Tfm(),_Tfm1()]], splits=[[1,2,5,7],[0,3,4,6]])\ndls = dsets.dataloaders(bs=4, device=torch.device('cpu'))", "_____no_output_____" ], [ "dsets = Datasets(range(8), [[_Tfm(),_Tfm1()]], splits=[[1,2,5,7],[0,3,4,6]])\ndls = dsets.dataloaders(bs=4, device=torch.device('cpu'))\ntst_dl = dls.test_dl([2,3,4,5])\ntest_eq(tst_dl._n_inp, 1)\ntest_eq(list(tst_dl), [(tensor([ 4, 6, 8, 10]),)])\n#Test you can change transforms\ntst_dl = dls.test_dl([2,3,4,5], after_item=add1)\ntest_eq(list(tst_dl), [(tensor([ 5, 7, 9, 11]),)])", "_____no_output_____" ] ], [ [ "## Export -", "_____no_output_____" ] ], [ [ "#hide\nfrom nbdev.export import notebook2script\nnotebook2script()", "Converted 00_torch_core.ipynb.\nConverted 01_layers.ipynb.\nConverted 01a_losses.ipynb.\nConverted 02_data.load.ipynb.\nConverted 03_data.core.ipynb.\nConverted 04_data.external.ipynb.\nConverted 05_data.transforms.ipynb.\nConverted 06_data.block.ipynb.\nConverted 07_vision.core.ipynb.\nConverted 08_vision.data.ipynb.\nConverted 09_vision.augment.ipynb.\nConverted 09b_vision.utils.ipynb.\nConverted 09c_vision.widgets.ipynb.\nConverted 10_tutorial.pets.ipynb.\nConverted 10b_tutorial.albumentations.ipynb.\nConverted 11_vision.models.xresnet.ipynb.\nConverted 12_optimizer.ipynb.\nConverted 13_callback.core.ipynb.\nConverted 13a_learner.ipynb.\nConverted 13b_metrics.ipynb.\nConverted 14_callback.schedule.ipynb.\nConverted 14a_callback.data.ipynb.\nConverted 15_callback.hook.ipynb.\nConverted 15a_vision.models.unet.ipynb.\nConverted 16_callback.progress.ipynb.\nConverted 17_callback.tracker.ipynb.\nConverted 18_callback.fp16.ipynb.\nConverted 18a_callback.training.ipynb.\nConverted 18b_callback.preds.ipynb.\nConverted 19_callback.mixup.ipynb.\nConverted 20_interpret.ipynb.\nConverted 20a_distributed.ipynb.\nConverted 21_vision.learner.ipynb.\nConverted 22_tutorial.imagenette.ipynb.\nConverted 23_tutorial.vision.ipynb.\nConverted 24_tutorial.siamese.ipynb.\nConverted 24_vision.gan.ipynb.\nConverted 30_text.core.ipynb.\nConverted 31_text.data.ipynb.\nConverted 32_text.models.awdlstm.ipynb.\nConverted 33_text.models.core.ipynb.\nConverted 34_callback.rnn.ipynb.\nConverted 35_tutorial.wikitext.ipynb.\nConverted 36_text.models.qrnn.ipynb.\nConverted 37_text.learner.ipynb.\nConverted 38_tutorial.text.ipynb.\nConverted 39_tutorial.transformers.ipynb.\nConverted 40_tabular.core.ipynb.\nConverted 41_tabular.data.ipynb.\nConverted 42_tabular.model.ipynb.\nConverted 43_tabular.learner.ipynb.\nConverted 44_tutorial.tabular.ipynb.\nConverted 45_collab.ipynb.\nConverted 46_tutorial.collab.ipynb.\nConverted 50_tutorial.datablock.ipynb.\nConverted 60_medical.imaging.ipynb.\nConverted 61_tutorial.medical_imaging.ipynb.\nConverted 65_medical.text.ipynb.\nConverted 70_callback.wandb.ipynb.\nConverted 71_callback.tensorboard.ipynb.\nConverted 72_callback.neptune.ipynb.\nConverted 73_callback.captum.ipynb.\nConverted 74_callback.cutmix.ipynb.\nConverted 97_test_utils.ipynb.\nConverted 99_pytorch_doc.ipynb.\nConverted dev-setup.ipynb.\nConverted index.ipynb.\nConverted quick_start.ipynb.\nConverted tutorial.ipynb.\n" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "<img src=\"images/logo.jpg\" style=\"display: block; margin-left: auto; margin-right: auto;\" alt=\"לוגו של מיזם לימוד הפייתון. נחש מצויר בצבעי צהוב וכחול, הנע בין האותיות של שם הקורס: לומדים פייתון. הסלוגן המופיע מעל לשם הקורס הוא מיזם חינמי ללימוד תכנות בעברית.\">", "_____no_output_____" ], [ "# <span style=\"text-align: right; direction: rtl; float: right;\">התנהגות של פונקציות</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n בפסקאות הקרובות נבחן פונקציות מזווית ראייה מעט שונה מהרגיל.<br>\n בואו נקפוץ ישירות למים!\n</p>", "_____no_output_____" ], [ "### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">שם של פונקציה</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n תכונה מעניינת שמתקיימת בפייתון היא שפונקציה היא ערך, בדיוק כמו כל ערך אחר.<br>\n נגדיר פונקציה שמעלה מספר בריבוע:\n</p>", "_____no_output_____" ] ], [ [ "def square(x):\n return x ** 2", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נוכל לבדוק מאיזה טיפוס הפונקציה (אנחנו לא קוראים לה עם סוגריים אחרי שמה – רק מציינים את שמה):\n</p>", "_____no_output_____" ] ], [ [ "type(square)", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n ואפילו לבצע השמה שלה למשתנה, כך ששם המשתנה החדש יצביע עליה:\n</p>", "_____no_output_____" ] ], [ [ "ribua = square\n\nprint(square(5))\nprint(ribua(5))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n מה מתרחש בתא למעלה?<br>\n כשהגדרנו את הפונקציה <var>square</var>, יצרנו לייזר עם התווית <var>square</var> שמצביע לפונקציה שמעלה מספר בריבוע.<br>\n בהשמה שביצענו בשורה הראשונה בתא שלמעלה, הלייזר שעליו מודבקת התווית <var>ribua</var> כוון אל אותה הפונקציה שעליה מצביע הלייזר <var>square</var>.<br>\n כעת <var>square</var> ו־<var>ribua</var> מצביעים לאותה פונקציה. אפשר לבדוק זאת כך:\n</p>", "_____no_output_____" ] ], [ [ "ribua is square", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n בשלב הזה אצטרך לבקש מכם לחגור חגורות, כי זה לא הולך להיות טיול רגיל הפעם.\n</p>", "_____no_output_____" ], [ "### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">פונקציות במבנים מורכבים</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n אם פונקציה היא בסך הכול ערך, ואם אפשר להתייחס לשם שלה בכל מקום, אין סיבה שלא נוכל ליצור רשימה של פונקציות!<br>\n ננסה לממש את הרעיון:\n</p>", "_____no_output_____" ] ], [ [ "def add(num1, num2):\n return num1 + num2\n\n\ndef subtract(num1, num2):\n return num1 - num2\n\n\ndef multiply(num1, num2):\n return num1 * num2\n\n\ndef divide(num1, num2):\n return num1 / num2\n\n\nfunctions = [add, subtract, multiply, divide]", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כעת יש לנו רשימה בעלת 4 איברים, שכל אחד מהם מצביע לפונקציה שונה.<br>\n אם נרצה לבצע פעולת חיבור, נוכל לקרוא ישירות ל־<var>add</var> או (בשביל התרגול) לנסות לאחזר אותה מהרשימה שיצרנו:\n</p>", "_____no_output_____" ] ], [ [ "# Option 1\nprint(add(5, 2))\n\n# Option 2\nmath_function = functions[0]\nprint(math_function(5, 2))\n\n# Option 3 (ugly, but works!)\nprint(functions[0](5, 2))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n אם נרצה, נוכל אפילו לעבור על רשימת הפונקציות בעזרת לולאה ולהפעיל את כולן, זו אחר זו:\n</p>", "_____no_output_____" ] ], [ [ "for function in functions:\n print(function(5, 2))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n בכל איטרציה של לולאת ה־<code>for</code>, המשתנה <var>function</var> עבר להצביע על הפונקציה הבאה מתוך רשימת <var>functions</var>.<br>\n בשורה הבאה קראנו לאותה הפונקציה ש־<var>function</var> מצביע עליה, והדפסנו את הערך שהיא החזירה. \n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כיוון שרשימה היא מבנה ששומר על סדר האיברים שבו, התוצאות מודפסות בסדר שבו הפונקציות שמורות ברשימה.<br>\n התוצאה הראשונה שאנחנו רואים היא תוצאת פונקציית החיבור, השנייה היא תוצאת פונקציית החיסור וכן הלאה.\n</p>", "_____no_output_____" ], [ "#### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">תרגיל ביניים: סוגרים חשבון</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כתבו פונקציה בשם <var>calc</var> שמקבלת כפרמטר שני מספרים וסימן של פעולה חשבונית.<br>\n הסימן יכול להיות אחד מאלה: <code>+</code>, <code>-</code>, <code>*</code> או <code>/</code>.<br>\n מטרת הפונקציה היא להחזיר את תוצאת הביטוי החשבוני שהופעל על שני המספרים.<br>\n בפתרונכם, השתמשו בהגדרת הפונקציות מלמעלה ובמילון.\n</p>", "_____no_output_____" ], [ "### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">העברת פונקציה כפרמטר</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נמשיך ללהטט בפונקציות.<br>\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n פונקציה נקראת \"<dfn>פונקציה מסדר גבוה</dfn>\" (<dfn>higher order function</dfn>) אם היא מקבלת כפרמטר פונקציה.<br>\n ניקח לדוגמה את הפונקציה <var>calculate</var>:\n</p>", "_____no_output_____" ] ], [ [ "def calculate(function, num1, num2):\n return function(num1, num2)", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n בקריאה ל־<var>calculate</var>, נצטרך להעביר פונקציה ושני מספרים.<br>\n נעביר לדוגמה את הפונקציה <var>divide</var> שהגדרנו קודם לכן:\n</p>", "_____no_output_____" ] ], [ [ "calculate(divide, 5, 2)", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n מה שמתרחש במקרה הזה הוא שהעברנו את הפונקציה <var>divide</var> כארגומנט ראשון.<br>\n הפרמטר <var>function</var> בפונקציה <var>calculate</var> מצביע כעת על פונקציית החילוק שהגדרנו למעלה.<br>\n מכאן, שהפונקציה תחזיר את התוצאה של <code>divide(5, 2)</code> – הרי היא 2.5.\n</p>", "_____no_output_____" ], [ "#### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">תרגיל ביניים: מפה לפה</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כתבו generator בשם <var>apply</var> שמקבל כפרמטר ראשון פונקציה (<var>func</var>), וכפרמטר שני iterable (<var dir=\"rtl\">iter</var>).<br>\n עבור כל איבר ב־iterable, ה־generator יניב את האיבר אחרי שהופעלה עליו הפונקציה <var>func</var>, דהיינו – <code>func(item)</code>.<br>\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n ודאו שהרצת התא הבא מחזירה <code>True</code> עבור הקוד שלכם:\n</p>", "_____no_output_____" ] ], [ [ "def square(number):\n return number ** 2\n\n\nsquare_check = apply(square, [5, -1, 6, -8, 0])\ntuple(square_check) == (25, 1, 36, 64, 0)", "_____no_output_____" ] ], [ [ "### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">סיכום ביניים</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n וואו. זה היה די משוגע.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n אז למעשה, פונקציות בפייתון הן ערך לכל דבר, כמו מחרוזות ומספרים!<br>\n אפשר לאחסן אותן במשתנים, לשלוח אותן כארגומנטים ולכלול אותם בתוך מבני נתונים מורכבים יותר.<br>\n אנשי התיאוריה של מדעי המחשב נתנו להתנהגות כזו שם: \"<dfn>אזרח ממדרגה ראשונה</dfn>\" (<dfn>first class citizen</dfn>).<br>\n אם כך, אפשר להגיד על פונקציות בפייתון שהן אזרחיות ממדרגה ראשונה.\n</p>", "_____no_output_____" ], [ "## <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">פונקציות מסדר גבוה בפייתון</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n החדשות הטובות הן שכבר עשינו היכרות קלה עם המונח פונקציות מסדר גבוה.<br>\n עכשיו, כשאנחנו יודעים שמדובר בפונקציות שמקבלות פונקציה כפרמטר, נתחיל ללכלך קצת את הידיים.<br>\n נציג כמה פונקציות פייתוניות מעניינות שכאלו:\n</p>", "_____no_output_____" ], [ "### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">הפונקציה map</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n הפונקציה <var>map</var> מקבלת פונקציה כפרמטר הראשון, ו־iterable כפרמטר השני.<br>\n <var>map</var> מפעילה את הפונקציה מהפרמטר הראשון על כל אחד מהאיברים שהועברו ב־iterable.<br>\n היא מחזירה iterator שמורכב מהערכים שחזרו מהפעלת הפונקציה.<br>\n</p>", "_____no_output_____" ], [ "\n<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n במילים אחרות, <var>map</var> יוצרת iterable חדש.<br>\n ה־iterable כולל את הערך שהוחזר מהפונקציה עבור כל איבר ב־<code>iterable</code> שהועבר.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n לדוגמה:\n</p>", "_____no_output_____" ] ], [ [ "squared_items = map(square, [1, 6, -1, 8, 0, 3, -3, 9, -8, 8, -7])\nprint(tuple(squared_items))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n הפונקציה קיבלה כארגומנט ראשון את הפונקציה <var>square</var> שהגדרנו למעלה, שמטרתה העלאת מספר בריבוע.<br>\n כארגומנט שני היא קיבלה את רשימת כל המספרים שאנחנו רוצים שהפונקציה תרוץ עליהם.<br>\n כשהעברנו ל־<var>map</var> את הארגומנטים הללו, <var>map</var> החזירה לנו ב־iterator (מבנה שאפשר לעבור עליו איבר־איבר) את התוצאה:<br>\n הריבוע, קרי החזקה השנייה, של כל אחד מהאיברים ברשימה שהועברה כארגומנט השני.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n למעשה, אפשר להגיד ש־<code>map</code> שקולה לפונקציה הבאה:\n</p>", "_____no_output_____" ] ], [ [ "def my_map(function, iterable):\n for item in iterable:\n yield function(item)", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n הנה דוגמה נוספת לשימוש ב־<var>map</var>:\n</p>", "_____no_output_____" ] ], [ [ "numbers = [(2, 4), (1, 4, 2), (1, 3, 5, 6, 2), (3, )]\nsums = map(sum, numbers)\nprint(tuple(sums))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n במקרה הזה, בכל מעבר, קיבלה הפונקציה <var>sum</var> איבר אחד מתוך הרשימה – tuple.<br>\n היא סכמה את האיברים של כל tuple שקיבלה, וכך החזירה לנו את הסכומים של כל ה־tuple־ים – זה אחרי זה.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n ודוגמה אחרונה:\n</p>", "_____no_output_____" ] ], [ [ "def add_one(number):\n return number + 1\n\n\nincremented = map(add_one, (1, 2, 3))\nprint(tuple(incremented))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n בדוגמה הזו יצרנו פונקציה משל עצמנו, ואותה העברנו ל־map.<br>\n מטרת דוגמה זו היא להדגיש שאין שוני בין העברת פונקציה שקיימת בפייתון לבין פונקציה שאנחנו יצרנו.\n</p>", "_____no_output_____" ], [ "<div class=\"align-center\" style=\"display: flex; text-align: right; direction: rtl; clear: both;\">\n <div style=\"display: flex; width: 10%; float: right; clear: both;\">\n <img src=\"images/exercise.svg\" style=\"height: 50px !important;\" alt=\"תרגול\"> \n </div>\n <div style=\"width: 70%\">\n <p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כתבו פונקציה שמקבלת רשימת מחרוזות של שתי מילים: שם פרטי ושם משפחה.<br>\n הפונקציה תשתמש ב־map כדי להחזיר מכולן רק את השם הפרטי.\n </p>\n </div>\n <div style=\"display: flex; width: 20%; border-right: 0.1rem solid #A5A5A5; padding: 1rem 2rem;\">\n <p style=\"text-align: center; direction: rtl; justify-content: center; align-items: center; clear: both;\">\n <strong>חשוב!</strong><br>\n פתרו לפני שתמשיכו!\n </p>\n </div>\n</div>", "_____no_output_____" ], [ "### <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">הפונקציה filter</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n הפונקציה <var>filter</var> מקבלת פונקציה כפרמטר ראשון, ו־iterable כפרמטר שני.<br>\n <var>filter</var> מפעילה על כל אחד מאיברי ה־iterable את הפונקציה, ומחזירה את האיבר אך ורק אם הערך שחזר מהפונקציה שקול ל־<code>True</code>.<br>\n אם ערך ההחזרה שקול ל־<code>False</code> – הערך \"יבלע\" ב־<var>filter</var> ולא יחזור ממנה.\n</p>", "_____no_output_____" ], [ "\n<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n במילים אחרות, <var>filter</var> יוצרת iterable חדש ומחזירה אותו.<br>\n ה־iterable כולל רק את האיברים שעבורם הפונקציה שהועברה החזירה ערך השקול ל־<code>True</code>.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נבנה, לדוגמה, פונקציה שמחזירה אם אדם הוא בגיר.<br>\n הפונקציה תקבל כפרמטר גיל, ותחזיר <code>True</code> כאשר הגיל שהועבר לה הוא לפחות 18, ו־<code>False</code> אחרת.\n</p>", "_____no_output_____" ] ], [ [ "def is_mature(age):\n return age >= 18", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נגדיר רשימת גילים, ונבקש מ־<var>filter</var> לסנן אותם לפי הפונקציה שהגדרנו:\n</p>", "_____no_output_____" ] ], [ [ "ages = [0, 1, 4, 10, 20, 35, 56, 84, 120]\nmature_ages = filter(is_mature, ages)\nprint(tuple(mature_ages))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כפי שלמדנו, <var>filter</var> מחזירה לנו רק גילים השווים ל־18 או גדולים ממנו.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נחדד שהפונקציה שאנחנו מעבירים ל־<var>filter</var> לא חייבת להחזיר בהכרח <code>True</code> או <code>False</code>.<br>\n הערך 0, לדוגמה, שקול ל־<code>False</code>, ולכן <var>filter</var> תסנן כל ערך שעבורו הפונקציה תחזיר 0: \n</p>", "_____no_output_____" ] ], [ [ "to_sum = [(1, -1), (2, 5), (5, -3, -2), (1, 2, 3)]\nsum_is_not_zero = filter(sum, to_sum)\nprint(tuple(sum_is_not_zero))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n בתא האחרון העברנו ל־<var>filter</var> את sum כפונקציה שאותה אנחנו רוצים להפעיל, ואת <var>to_sum</var> כאיברים שעליהם אנחנו רוצים לפעול.<br>\n ה־tuple־ים שסכום איבריהם היה 0 סוננו, וקיבלנו חזרה iterator שהאיברים בו הם אך ורק אלו שסכומם שונה מ־0.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כטריק אחרון, נלמד ש־<var>filter</var> יכולה לקבל גם <code>None</code> בתור הפרמטר הראשון, במקום פונקציה.<br>\n זה יגרום ל־<var>filter</var> לא להפעיל פונקציה על האיברים שהועברו, כלומר לסנן אותם כמו שהם.<br>\n איברים השקולים ל־<code>True</code> יוחזרו, ואיברים השקולים ל־<code>False</code> לא יוחזרו:\n</p>", "_____no_output_____" ] ], [ [ "to_sum = [0, \"\", None, 0.0, True, False, \"Hello\"]\nequivalent_to_true = filter(None, to_sum)\nprint(tuple(equivalent_to_true))", "_____no_output_____" ] ], [ [ "<div class=\"align-center\" style=\"display: flex; text-align: right; direction: rtl; clear: both;\">\n <div style=\"display: flex; width: 10%; float: right; clear: both;\">\n <img src=\"images/exercise.svg\" style=\"height: 50px !important;\" alt=\"תרגול\"> \n </div>\n <div style=\"width: 70%\">\n <p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כתבו פונקציה שמקבלת רשימת מחרוזות, ומחזירה רק את המחרוזות הפלינדרומיות שבה.<br>\n מחרוזת נחשבת פלינדרום אם קריאתה מימין לשמאל ומשמאל לימין יוצרת אותו ביטוי.<br>\n השתמשו ב־<var>filter</var>.\n </p>\n </div>\n <div style=\"display: flex; width: 20%; border-right: 0.1rem solid #A5A5A5; padding: 1rem 2rem;\">\n <p style=\"text-align: center; direction: rtl; justify-content: center; align-items: center; clear: both;\">\n <strong>חשוב!</strong><br>\n פתרו לפני שתמשיכו!\n </p>\n </div>\n</div>", "_____no_output_____" ], [ "## <span style=\"text-align: right; direction: rtl; float: right; clear: both;\">פונקציות אנונימיות</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n תעלול נוסף שנוסיף לארגז הכלים שלנו הוא <dfn>פונקציות אנונימיות</dfn> (<dfn>anonymous functions</dfn>).<br>\n אל תיבהלו מהשם המאיים – בסך הכול פירושו הוא \"פונקציות שאין להן שם\".<br>\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n לפני שאתם מרימים גבה ושואלים את עצמכם למה הן שימושיות, בואו נבחן כמה דוגמאות.<br>\n ניזכר בהגדרת פונקציית החיבור שיצרנו לא מזמן:\n</p>", "_____no_output_____" ] ], [ [ "def add(num1, num2):\n return num1 + num2", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n ונגדיר את אותה הפונקציה בדיוק בצורה אנונימית:\n</p>", "_____no_output_____" ] ], [ [ "add = lambda num1, num2: num1 + num2\n\nprint(add(5, 2))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n לפני שנסביר איפה החלק של ה\"פונקציה בלי שם\" נתמקד בצד ימין של ההשמה.<br>\n כיצד מנוסחת הגדרת פונקציה אנונימית?\n</p>", "_____no_output_____" ], [ "<ol style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n <li>הצהרנו שברצוננו ליצור פונקציה אנונימית בעזרת מילת המפתח <code>lambda</code>.</li>\n <li>מייד אחריה, ציינו את שמות כל הפרמטרים שהפונקציה תקבל, כשהם מופרדים בפסיק זה מזה.</li>\n <li>כדי להפריד בין רשימת הפרמטרים לערך ההחזרה של הפונקציה, השתמשנו בנקודתיים.</li>\n <li>אחרי הנקודתיים, כתבנו את הביטוי שאנחנו רוצים שהפונקציה תחזיר.</li>\n</ol>", "_____no_output_____" ], [ "<figure>\n <img src=\"images/lambda.png\" style=\"max-width: 500px; margin-right: auto; margin-left: auto; text-align: center;\" alt=\"בתמונה מופיעה הגדרת ה־lambda שביצענו קודם לכן. מעל המילה lambda המודגשת בירוק ישנו פס מקווקו, ומעליו רשום 'הצהרה'. מימין למילה lambda כתוב num1 (פסיק) num2, מעליהם קו מקווקו ומעליו המילה 'פרמטרים'. מימין לפרמטרים יש נקודתיים, ואז num1 (הסימן פלוס) num2. מעליהם קו מקווקו, ומעליו המילה 'ערך החזרה'.\"/>\n <figcaption style=\"margin-top: 2rem; text-align: center; direction: rtl;\">חלקי ההגדרה של פונקציה אנונימית בעזרת מילת המפתח <code>lambda</code><br><span style=\"color: white;\">A girl has no name</span></figcaption>\n</figure>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n במה שונה ההגדרה של פונקציה זו מההגדרה של פונקציה רגילה?<br>\n היא לא באמת שונה.<br>\n המטרה היא לאפשר תחביר שיקל על חיינו כשאנחנו רוצים לכתוב פונקציה קצרצרה שאורכה שורה אחת.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נראה, לדוגמה, שימוש ב־<var>filter</var> כדי לסנן את כל האיברים שאינם חיוביים:\n</p>", "_____no_output_____" ] ], [ [ "def is_positive(number):\n return number > 0\n\n\nnumbers = [-2, -1, 0, 1, 2]\npositive_numbers = filter(is_positive, numbers)\nprint(tuple(positive_numbers))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n במקום להגדיר פונקציה חדשה שנקראת <var>is_positive</var>, נוכל להשתמש בפונקציה אנונימית:\n</p>", "_____no_output_____" ] ], [ [ "numbers = [-2, -1, 0, 1, 2]\npositive_numbers = filter(lambda n: n > 0, numbers)\nprint(tuple(positive_numbers))", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n איך זה עובד?<br>\n במקום להעביר ל־<var>filter</var> פונקציה שיצרנו מבעוד מועד, השתמשנו ב־<code>lambda</code> כדי ליצור פונקציה ממש באותה השורה.<br>\n הפונקציה שהגדרנו מקבלת מספר (<var>n</var>), ומחזירה <code>True</code> אם הוא חיובי, או <code>False</code> אחרת.<br>\n שימו לב שבצורה זו באמת לא היינו צריכים לתת שם לפונקציה שהגדרנו.\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n השימוש בפונקציות אנונימיות לא מוגבל ל־<var>map</var> ול־<var>filter</var>, כמובן.<br>\n מקובל להשתמש ב־<code>lambda</code> גם עבור פונקציות כמו <var>sorted</var>, שמקבלות פונקציה בתור ארגומנט.\n</p>", "_____no_output_____" ], [ "<div class=\"align-center\" style=\"display: flex; text-align: right; direction: rtl;\">\n <div style=\"display: flex; width: 10%; float: right; \">\n <img src=\"images/recall.svg\" style=\"height: 50px !important;\" alt=\"תזכורת\" title=\"תזכורת\"> \n </div>\n <div style=\"width: 90%\">\n <p style=\"text-align: right; direction: rtl;\">\n הפונקציה <code>sorted</code> מאפשרת לנו לסדר ערכים, ואפילו להגדיר עבורה לפי מה לסדר אותם.<br>\n לרענון בנוגע לשימוש בפונקציה גשו למחברת בנושא פונקציות מובנות בשבוע 4.\n </p>\n </div>\n</div>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נסדר, למשל, את הדמויות ברשימה הבאה, לפי תאריך הולדתן:\n</p>", "_____no_output_____" ] ], [ [ "closet = [\n {'name': 'Peter', 'year_of_birth': 1927, 'gender': 'Male'},\n {'name': 'Edmund', 'year_of_birth': 1930, 'gender': 'Male'},\n {'name': 'Lucy', 'year_of_birth': 1932, 'gender': 'Female'},\n {'name': 'Susan', 'year_of_birth': 1928, 'gender': 'Female'},\n {'name': 'Jadis', 'year_of_birth': 0, 'gender': 'Female'},\n]", "_____no_output_____" ] ], [ [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n נרצה שסידור הרשימה יתבצע לפי המפתח <var>year_of_birth</var>.<br>\n כלומר, בהינתן מילון שמייצג דמות בשם <var>d</var>, יש להשיג את <code dir=\"ltr\">d['year_of_birth']</code>, ולפיו לבצע את סידור הרשימה.<br>\n ניגש למלאכה:\n</p>", "_____no_output_____" ] ], [ [ "sorted(closet, key=lambda d: d['year_of_birth'])", "_____no_output_____" ] ], [ [ "<div class=\"align-center\" style=\"display: flex; text-align: right; direction: rtl;\">\n <div style=\"display: flex; width: 10%; float: right; \">\n <img src=\"images/warning.png\" style=\"height: 50px !important;\" alt=\"אזהרה!\"> \n </div>\n <div style=\"width: 90%\">\n <p style=\"text-align: right; direction: rtl;\">\n פונקציות אנונימיות הן יכולת חביבה שאמורה לסייע לכם לכתוב קוד נאה וקריא.<br>\n כלל אצבע טוב לחיים הוא להימנע משימוש בהן כאשר הן מסרבלות את הקוד.\n </p>\n </div>\n</div>", "_____no_output_____" ], [ "<div class=\"align-center\" style=\"display: flex; text-align: right; direction: rtl; clear: both;\">\n <div style=\"display: flex; width: 10%; float: right; clear: both;\">\n <img src=\"images/exercise.svg\" style=\"height: 50px !important;\" alt=\"תרגיל\" title=\"תרגיל\"> \n </div>\n <div style=\"width: 70%\">\n <p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n סדרו את הדמויות ב־<var>closet</var> לפי האות האחרונה בשמם. \n </p>\n </div>\n</div>", "_____no_output_____" ], [ "## <span style=\"align: right; direction: rtl; float: right; clear: both;\">מונחים</span>", "_____no_output_____" ], [ "<dl style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n <dt>פונקציה מסדר גבוה</dt>\n <dd>פונקציה שמקבלת פונקציה כאחד הארגומנטים, או שמחזירה פונקציה כערך ההחזרה שלה.</dd>\n <dt>אזרח ממדרגה ראשונה</dt>\n <dd>ישות תכנותית המתנהגת בשפת התכנות כערך לכל דבר. בפייתון, פונקציות הן אזרחיות ממדרגה ראשונה.<dd>\n <dt>פונקציה אנונימית, פונקציית <code>lambda</code></dt>\n <dd>פונקציה ללא שם המיועדת להגדרת פונקציה בשורה אחת, לרוב לשימוש חד־פעמי. בעברית: פונקציית למדא.</dd>\n</dl>", "_____no_output_____" ], [ "## <span style=\"align: right; direction: rtl; float: right; clear: both;\">תרגילים</span>", "_____no_output_____" ], [ "### <span style=\"align: right; direction: rtl; float: right; clear: both;\">פילטר מותאם אישית</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כתבו פונקציה בשם <var>my_filter</var> שמתנהגת בדיוק כמו הפונקציה <var>filter</var>.<br>\n בפתירת התרגיל, המנעו משימוש ב־<var>filter</var> או במודולים.\n</p>", "_____no_output_____" ], [ "### <span style=\"align: right; direction: rtl; float: right; clear: both;\">נשאר? חיובי</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\nכתבו פונקציה בשם <var>get_positive_numbers</var> שמקבלת מהמשתמש קלט בעזרת <var>input</var>.<br>\nהמשתמש יתבקש להזין סדרה של מספרים המופרדים בפסיק זה מזה.<br>\nהפונקציה תחזיר את כל המספרים החיוביים שהמשתמש הזין, כרשימה של מספרים מסוג <code>int</code>.<br>\nאפשר להניח שהקלט מהמשתמש תקין.\n</p>", "_____no_output_____" ], [ "### <span style=\"align: right; direction: rtl; float: right; clear: both;\">ריצת 2,000</span>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n כתבו פונקציה בשם <var>timer</var> שמקבלת כפרמטר פונקציה (נקרא לה <var>f</var>) ופרמטרים נוספים.<br>\n הפונקציה <var>timer</var> תמדוד כמה זמן רצה פונקציה <var>f</var> כשמועברים אליה אותם פרמטרים. <br>\n</p>", "_____no_output_____" ], [ "<p style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n לדוגמה:\n</p>", "_____no_output_____" ], [ "<ol style=\"text-align: right; direction: rtl; float: right; clear: both;\">\n <li>עבור הקריאה <code dir=\"ltr\">timer(print, \"Hello\")</code>, תחזיר הפונקציה את משך זמן הביצוע של <code dir=\"ltr\">print(\"Hello\")</code>.</li>\n <li>עבור הקריאה <code dir=\"ltr\">timer(zip, [1, 2, 3], [4, 5, 6])</code>, תחזיר הפונקציה את משך זמן הביצוע של <code dir=\"ltr\">zip([1, 2, 3], [4, 5, 6])</code>.</li>\n <li>עבור הקריאה <code dir=\"ltr\">timer(\"Hi {name}\".format, name=\"Bug\")</code>, תחזיר הפונקציה את משך זמן הביצוע של <code dir=\"ltr\">\"Hi {name}\".format(name=\"Bug\")</code></li>\n</ol>", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "BSD-2-Clause" ]

[ "BSD-2-Clause" ]

[ "BSD-2-Clause" ]

[ [ [ "import os\nimport json\nimport pickle\nimport random\nfrom collections import defaultdict, Counter\n\nfrom indra.literature.adeft_tools import universal_extract_text\nfrom indra.databases.hgnc_client import get_hgnc_name, get_hgnc_id\n\nfrom adeft.discover import AdeftMiner\nfrom adeft.gui import ground_with_gui\nfrom adeft.modeling.label import AdeftLabeler\nfrom adeft.modeling.classify import AdeftClassifier\nfrom adeft.disambiguate import AdeftDisambiguator, load_disambiguator\n\n\nfrom adeft_indra.ground.ground import AdeftGrounder\nfrom adeft_indra.model_building.s3 import model_to_s3\nfrom adeft_indra.model_building.escape import escape_filename\nfrom adeft_indra.db.content import get_pmids_for_agent_text, get_pmids_for_entity, \\\n get_plaintexts_for_pmids", "_____no_output_____" ], [ "adeft_grounder = AdeftGrounder()", "_____no_output_____" ], [ "shortforms = ['UBC']\nmodel_name = ':'.join(sorted(escape_filename(shortform) for shortform in shortforms))\nresults_path = os.path.abspath(os.path.join('../..', 'results', model_name))", "_____no_output_____" ], [ "miners = dict()\nall_texts = {}\nfor shortform in shortforms:\n pmids = get_pmids_for_agent_text(shortform)\n text_dict = get_plaintexts_for_pmids(pmids, contains=shortforms)\n text_dict = {pmid: text for pmid, text in text_dict.items() if len(text) > 5}\n miners[shortform] = AdeftMiner(shortform)\n miners[shortform].process_texts(text_dict.values())\n all_texts.update(text_dict)\n\nlongform_dict = {}\nfor shortform in shortforms:\n longforms = miners[shortform].get_longforms()\n longforms = [(longform, count, score) for longform, count, score in longforms\n if count*score > 2]\n longform_dict[shortform] = longforms\n \ncombined_longforms = Counter()\nfor longform_rows in longform_dict.values():\n combined_longforms.update({longform: count for longform, count, score\n in longform_rows})\ngrounding_map = {}\nnames = {}\nfor longform in combined_longforms:\n groundings = adeft_grounder.ground(longform)\n if groundings:\n grounding = groundings[0]['grounding']\n grounding_map[longform] = grounding\n names[grounding] = groundings[0]['name']\nlongforms, counts = zip(*combined_longforms.most_common())\npos_labels = []", "_____no_output_____" ], [ "list(zip(longforms, counts))", "_____no_output_____" ], [ "try:\n disamb = load_disambiguator(shortforms[0])\n for shortform, gm in disamb.grounding_dict.items():\n for longform, grounding in gm.items():\n grounding_map[longform] = grounding\n for grounding, name in disamb.names.items():\n names[grounding] = name\n pos_labels = disamb.pos_labels\nexcept Exception:\n pass", "_____no_output_____" ], [ "names", "_____no_output_____" ], [ "grounding_map, names, pos_labels = ground_with_gui(longforms, counts, \n grounding_map=grounding_map,\n names=names, pos_labels=pos_labels, no_browser=True, port=8891)", "INFO: [2020-11-10 03:01:29] indra.ontology.bio.ontology - Loading INDRA bio ontology from cache at /home/ubuntu/.indra/bio_ontology/1.4/bio_ontology.pkl\n" ], [ "result = [grounding_map, names, pos_labels]", "_____no_output_____" ], [ "result", "_____no_output_____" ], [ "grounding_map, names, pos_labels = [{'ubiquitin c': 'HGNC:12468',\n 'ubiquitin conjugating': 'FPLX:UBE2',\n 'ubiquitin conjugating enzyme': 'FPLX:UBE2',\n 'unipolar brush cells': 'ungrounded',\n 'urinary bladder cancer': 'MESH:D001749',\n 'urothelial bladder cancer': 'MESH:D001749',\n 'urothelial bladder carcinoma': 'MESH:D001749'},\n {'HGNC:12468': 'UBC',\n 'FPLX:UBE2': 'UBE2',\n 'MESH:D001749': 'Urinary Bladder Neoplasms'},\n ['FPLX:UBE2', 'HGNC:12468', 'MESH:D001749']]", "_____no_output_____" ], [ "excluded_longforms = []", "_____no_output_____" ], [ "grounding_dict = {shortform: {longform: grounding_map[longform] \n for longform, _, _ in longforms if longform in grounding_map\n and longform not in excluded_longforms}\n for shortform, longforms in longform_dict.items()}\nresult = [grounding_dict, names, pos_labels]\n\nif not os.path.exists(results_path):\n os.mkdir(results_path)\nwith open(os.path.join(results_path, f'{model_name}_preliminary_grounding_info.json'), 'w') as f:\n json.dump(result, f)", "_____no_output_____" ], [ "additional_entities = {}", "_____no_output_____" ], [ "unambiguous_agent_texts = {}", "_____no_output_____" ], [ "labeler = AdeftLabeler(grounding_dict)\ncorpus = labeler.build_from_texts((text, pmid) for pmid, text in all_texts.items())\nagent_text_pmid_map = defaultdict(list)\nfor text, label, id_ in corpus:\n agent_text_pmid_map[label].append(id_)\n\nentity_pmid_map = {entity: set(get_pmids_for_entity(*entity.split(':', maxsplit=1),\n major_topic=True))for entity in additional_entities}", "_____no_output_____" ], [ "intersection1 = []\nfor entity1, pmids1 in entity_pmid_map.items():\n for entity2, pmids2 in entity_pmid_map.items():\n intersection1.append((entity1, entity2, len(pmids1 & pmids2)))", "_____no_output_____" ], [ "intersection2 = []\nfor entity1, pmids1 in agent_text_pmid_map.items():\n for entity2, pmids2 in entity_pmid_map.items():\n intersection2.append((entity1, entity2, len(set(pmids1) & pmids2)))", "_____no_output_____" ], [ "intersection1", "_____no_output_____" ], [ "intersection2", "_____no_output_____" ], [ "all_used_pmids = set()\nfor entity, agent_texts in unambiguous_agent_texts.items():\n used_pmids = set()\n for agent_text in agent_texts[1]:\n pmids = set(get_pmids_for_agent_text(agent_text))\n new_pmids = list(pmids - all_texts.keys() - used_pmids)\n text_dict = get_plaintexts_for_pmids(new_pmids, contains=agent_texts)\n corpus.extend([(text, entity, pmid) for pmid, text in text_dict.items() if len(text) >= 5])\n used_pmids.update(new_pmids)\n all_used_pmids.update(used_pmids)\n \nfor entity, pmids in entity_pmid_map.items():\n new_pmids = list(set(pmids) - all_texts.keys() - all_used_pmids)\n if len(new_pmids) > 10000:\n new_pmids = random.choices(new_pmids, k=10000)\n _, contains = additional_entities[entity]\n text_dict = get_plaintexts_for_pmids(new_pmids, contains=contains)\n corpus.extend([(text, entity, pmid) for pmid, text in text_dict.items() if len(text) >= 5])", "_____no_output_____" ], [ "names.update({key: value[0] for key, value in additional_entities.items()})\nnames.update({key: value[0] for key, value in unambiguous_agent_texts.items()})\npos_labels = list(set(pos_labels) | additional_entities.keys() |\n unambiguous_agent_texts.keys())", "_____no_output_____" ], [ "%%capture\n\nclassifier = AdeftClassifier(shortforms, pos_labels=pos_labels, random_state=1729)\nparam_grid = {'C': [100.0], 'max_features': [10000]}\ntexts, labels, pmids = zip(*corpus)\nclassifier.cv(texts, labels, param_grid, cv=5, n_jobs=5)", "INFO: [2020-11-10 03:02:57] /adeft/PP/adeft/adeft/modeling/classify.py - Beginning grid search in parameter space:\n{'C': [100.0], 'max_features': [10000]}\nINFO: [2020-11-10 03:02:59] /adeft/PP/adeft/adeft/modeling/classify.py - Best f1 score of 0.8145351127630539 found for parameter values:\n{'logit__C': 100.0, 'tfidf__max_features': 10000}\n" ], [ "classifier.stats", "_____no_output_____" ], [ "disamb = AdeftDisambiguator(classifier, grounding_dict, names)", "_____no_output_____" ], [ "disamb.dump(model_name, results_path)", "_____no_output_____" ], [ "print(disamb.info())", "Disambiguation model for UBC\n\nProduces the disambiguations:\n\tUBC*\tHGNC:12468\n\tUBE2*\tFPLX:UBE2\n\tUrinary Bladder Neoplasms*\tMESH:D001749\n\nClass level metrics:\n--------------------\nGrounding \tCount\tF1 \nUrinary Bladder Neoplasms*\t57\t0.91538\n UBC*\t16\t0.55778\n UBE2*\t10\t0.62667\n Ungrounded\t 2\t 0.0\n\nWeighted Metrics:\n-----------------\n\tF1 score:\t0.81454\n\tPrecision:\t0.8339\n\tRecall:\t\t0.85515\n\n* Positive labels\nSee Docstring for explanation\n\n" ], [ "model_to_s3(disamb)", "_____no_output_____" ], [ "preds = [disamb.disambiguate(text) for text in all_texts.values()]", "_____no_output_____" ], [ "texts = [text for pred, text in zip(preds, all_texts.values()) if pred[0] == 'HGNC:10967']", "_____no_output_____" ], [ "texts[3]", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Unlicense" ]

[ "Unlicense" ]

[ "Unlicense" ]

[ [ [ "# Teoremas Asintóticos\n\n## Ley débil de los grandes números\nSea $\\{X_i\\}_{i=1}^N$ un conjunto de $N$ v.a. independientes idénticamente distribuidas (iid) de media $\\mu$ y varianza finita $\\sigma^2$. Entonces, para cualquier $\\epsilon >0$ se tiene:\n\n$$\\lim_{N \\to \\infty} P\\left\\{\\middle | \\frac{\\sum_i X_i}{N} - \\mu \\,\\middle | > \\epsilon \\right\\} = 0$$\n", "_____no_output_____" ], [ "## Teorema del Límite Central\nSean $X_1,...X_N$ v.a. iid de media $\\mu$ y varianza $\\sigma^2$ entonces\n$S_N=X_1+...X_N$ cumple:\n\n$$ \\lim_{N \\to \\infty}P\\left( \\frac{S_N - N\\mu}{\\sigma\\sqrt N} \\leq z\\right) = F_Z(z) \n\\qquad Z \\sim \\cal{N}(0,1)$$\n\n\nEn particular $m= \\frac{S_N}{N}= \\frac{\\sum_{i=1}^N \\limits X_i}{N}$ cumple:\n\n$$ \\lim_{N \\to \\infty}P\\left( \\frac{m - \\mu}{\\sigma/\\sqrt N} \\leq z\\right) = F_Z(z) $$\n\nEs decir que es posible aproximar la distribución de $m$ por ${\\cal{N}}(\\mu,\\frac{\\sigma^2}{N})$\n", "_____no_output_____" ] ] ]

[ "markdown" ]

[ [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Identificar Perfil de Consumo de Clientes de uma Instituição Financeira", "_____no_output_____" ], [ "## Uma breve introdução", "_____no_output_____" ], [ "Uma Instituição Financeira X tem o interesse em identificar o perfil de gastos dos seus clientes. Identificando os clientes certos, eles podem melhorar a comunicação dos ativos promocionais, utilizar com mais eficiência os canais de comunicação e aumentar o engajamento dos clientes com o uso do seu produto.", "_____no_output_____" ], [ "## Sobre o estudo", "_____no_output_____" ], [ "Os dados estão anonimizados por questões de segurança. No dataset, temos o valor gasto de 121818 clientes, no ano de 2019, em cada ramo de atividade. A base já está limpa. O intuito aqui é apresentar uma possibilidade de fazer a clusterização de clientes com base em seu consumo. Os dados são de 1 ano a fim de diminuir o efeito da sazonalidade.", "_____no_output_____" ], [ "## Importando a biblioteca", "_____no_output_____" ], [ "Será necessário instalar o pacote kmodes, caso não tenha. ", "_____no_output_____" ] ], [ [ "# pip install --upgrade kmodes", "_____no_output_____" ], [ "import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nimport os\nimport random\nimport plotly.express as px\n# pacote do modelo\nfrom kmodes.kmodes import KModes\n# from sklearn.model_selection import train_test_split\n\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\nrandom.seed(2020)\n\npd.options.display.float_format = '{:,.2f}'.format", "_____no_output_____" ] ], [ [ "## Leitura dos dados", "_____no_output_____" ] ], [ [ "# Carregar os dados\ndados = pd.read_csv('../../dados_gasto_rmat.csv', sep=',')", "_____no_output_____" ], [ "dados.head(10)", "_____no_output_____" ], [ "print('O arquivo possui ' + str(dados.shape[0]) + ' linhas e ' + str(dados.shape[1]) + ' colunas.')", "O arquivo possui 3898614 linhas e 3 colunas.\n" ], [ "# Tipos dos dados\ndados.dtypes", "_____no_output_____" ], [ "# Verificando valores nulos\ndados.isnull().values.any()", "_____no_output_____" ] ], [ [ "A base não possui valores nulos.", "_____no_output_____" ], [ "## Visualização dos dados", "_____no_output_____" ] ], [ [ "# Total de clientes\nlen(dados['CLIENTE'].unique())", "_____no_output_____" ] ], [ [ "Soma do valor gasto em cada ramo de atividade.", "_____no_output_____" ] ], [ [ "# Dados agregados por RMAT e o percentual gasto\ndados_agreg = dados.groupby(['RMAT'])['VALOR_GASTO'].sum().reset_index()\ndados_agreg['percentual'] = round(dados_agreg['VALOR_GASTO']/sum(dados_agreg['VALOR_GASTO'])*100,2)\ndados_agreg.head()", "_____no_output_____" ], [ "# Traz o 30 RMATs que tiveram maior gasto dos clientes\ntop_rotulos = dados_agreg.sort_values(by = 'percentual', ascending = False)[:30]", "_____no_output_____" ], [ "# Gráfico dos RMATs\nax = top_rotulos.plot.barh(x='RMAT', y='percentual', rot=0, figsize = (20, 15), fontsize=20, color='violet')\nplt.title('Percentual do valor gasto por Ramo de Atividade', fontsize=22)\nplt.xlabel('')\nplt.ylabel('')\nplt.show()", "_____no_output_____" ] ], [ [ "Supermercados, postos de combustível e drogarias são os ramos com maior consumo dos clientes, representando 13.48%, 7.07% e 5.75%, respectivamente. ", "_____no_output_____" ], [ "### Construção da base para o modelo", "_____no_output_____" ], [ "Vamos construir uma base que seja a nível cliente e as colunas serão os percentuais de consumo em cada ramo de atividade. Aqui, temos muitos percentuais 0 em determinados ramos de atividade e isso tem impacto negativo na construção do modelo. Portanto, iremos categorizar as variáveis de acordo com a variação do percentual para cada atributo.", "_____no_output_____" ] ], [ [ "dados.head()", "_____no_output_____" ] ], [ [ "A função pivot faz a transposição dos dados, transformando o valor de \"RMAT\" em colunas.", "_____no_output_____" ] ], [ [ "# Inverter o data frame (colocar nos rmats como coluna)\ncli_pivot = dados.pivot(index='CLIENTE', columns='RMAT', values='VALOR_GASTO')\ncli_pivot.fillna(0, inplace = True)", "_____no_output_____" ], [ "# Calcular o percentual de volume transacionado de cada cliente por rmat \ncli_pivot = cli_pivot.apply(lambda x: x.apply(lambda y: 100*y/sum(x)),axis = 1)\ncli_pivot.head()", "_____no_output_____" ] ], [ [ "Abaixo, uma função que quebra os valores de cada coluna em quantidade de categorias escolhido. Nesse caso, utilizamos 8 quebras.", "_____no_output_____" ] ], [ [ "# Funnção para fazer categorização das variaveis\ndef hcut(df, colunas, nlevs, prefixo=''):\n x = df.copy()\n for c in colunas:\n \n x[prefixo+c] = pd.cut(x[c] , bins=nlevs, include_lowest = False, precision=0)\n \n return x", "_____no_output_____" ], [ "base_cluster = hcut(cli_pivot, cli_pivot\n .columns, 8, 'esc_')", "_____no_output_____" ], [ "base_cluster.head()", "_____no_output_____" ] ], [ [ "Agora, filtraremos as variáveis necessárias para a modelagem. ", "_____no_output_____" ] ], [ [ "# Selecionar somentes as colunas categorizadas que serão utilizadas no modelo\nfilter_col = [col for col in base_cluster if\n col.startswith('esc_')]\ndf1 = base_cluster.loc[:,filter_col].reset_index() \ndf1.head()", "_____no_output_____" ] ], [ [ "### Criação do modelo ", "_____no_output_____" ], [ "Para nosso caso, utilizaremos o método de clusterização chamado K-modes. Esse método é uma extensão do K-means. Em vez de distâncias, ele usa dissimilaridade (isto é, quantificação das incompatibilidades totais entre dois objetos: quanto menor esse número, masi semelhantes são os dois objetos). Além disso, ele utiliza modas. Cada vetor de elementos é criado de forma a minimizar as diferenças entre o próprio vetor e cada objetos dos dados. Assim, teremos tantos vetores de modas quanto o número de clusters necessários, pois eles atuam como centróides. \n", "_____no_output_____" ], [ "Aqui, iremos fazer a divisão dos clientes em 7 clusters. ", "_____no_output_____" ] ], [ [ "km_huang = KModes(n_clusters=7, init = \"Huang\", n_init = 5, verbose=1, random_state=2020)\nfitClusters = km_huang.fit_predict(df1)", "Init: initializing centroids\nInit: initializing clusters\nStarting iterations...\nRun 1, iteration: 1/100, moves: 37911, cost: 285526.0\nRun 1, iteration: 2/100, moves: 6829, cost: 285526.0\nInit: initializing centroids\nInit: initializing clusters\nStarting iterations...\nRun 2, iteration: 1/100, moves: 24674, cost: 296153.0\nInit: initializing centroids\nInit: initializing clusters\nStarting iterations...\nRun 3, iteration: 1/100, moves: 38186, cost: 311798.0\nRun 3, iteration: 2/100, moves: 1175, cost: 311797.0\nRun 3, iteration: 3/100, moves: 0, cost: 311797.0\nInit: initializing centroids\nInit: initializing clusters\nStarting iterations...\nRun 4, iteration: 1/100, moves: 49330, cost: 291921.0\nRun 4, iteration: 2/100, moves: 5319, cost: 291920.0\nRun 4, iteration: 3/100, moves: 0, cost: 291920.0\nInit: initializing centroids\nInit: initializing clusters\nStarting iterations...\nRun 5, iteration: 1/100, moves: 35294, cost: 289461.0\nRun 5, iteration: 2/100, moves: 14295, cost: 289460.0\nRun 5, iteration: 3/100, moves: 0, cost: 289460.0\nBest run was number 1\n" ], [ "# Adiciona os clusters na base\ndf1['cluster'] = fitClusters\nbase_cluster['cluster'] = fitClusters", "_____no_output_____" ] ], [ [ "#### Percentual de clientes por cluster. ", "_____no_output_____" ] ], [ [ "df1['cluster'].value_counts()/len(df1)*100", "_____no_output_____" ], [ "df2 = base_cluster.drop(columns=filter_col)", "_____no_output_____" ], [ "df2.head()", "_____no_output_____" ], [ "# Para visualização dos clusters\n\n# from sklearn.decomposition import PCA\n# pca_2 = PCA(2)\n# plot_columns = pca_2.fit_transform(base_cluster.iloc[:,0:65])\n# plt.scatter(x=plot_columns[:,0], y=plot_columns[:,1], c=fitClusters,)\n# plt.show()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ] ]

[ "DOC" ]

[ "DOC" ]

[ "DOC" ]

[ [ [ "# Iterators and Generators Homework \n\n### Problem 1\n\nCreate a generator that generates the squares of numbers up to some number N.", "_____no_output_____" ] ], [ [ "# option 1\ndef gensquares(N):\n end = N\n start = 0\n while start<end:\n yield start**2\n start += 1\n\n# option 2\ndef gensquares2(N):\n\n for num in range(N):\n yield num**2\n\n\n ", "_____no_output_____" ], [ "for x in gensquares2(10):\n print(x)", "0\n1\n4\n9\n16\n25\n36\n49\n64\n81\n" ] ], [ [ "### Problem 2\n\nCreate a generator that yields \"n\" random numbers between a low and high number (that are inputs). <br>Note: Use the random library. For example:", "_____no_output_____" ] ], [ [ "import random\n\nrandom.randint(1,10)", "_____no_output_____" ], [ "def rand_num(low,high,n):\n for times in range(n):\n yield random.randint(low,high)\n\n ", "_____no_output_____" ], [ "for num in rand_num(1,10,12):\n print(num)", "1\n5\n4\n6\n9\n5\n5\n4\n2\n7\n9\n6\n" ] ], [ [ "### Problem 3\n\nUse the iter() function to convert the string below into an iterator:\n", "_____no_output_____" ] ], [ [ "s = 'hello'\n\n#code here\ns_iter = iter()", "_____no_output_____" ] ], [ [ "### Problem 4\nExplain a use case for a generator using a yield statement where you would not want to use a normal function with a return statement.<br><br><br><br><br><br>\n\n", "_____no_output_____" ], [ "### Extra Credit!\nCan you explain what *gencomp* is in the code below? (Note: We never covered this in lecture! You will have to do some Googling/Stack Overflowing!)", "_____no_output_____" ] ], [ [ "my_list = [1,2,3,4,5]\n\ngencomp = (item for item in my_list if item > 3)\n\nfor item in gencomp:\n print(item)", "4\n5\n" ] ], [ [ "Hint: Google *generator comprehension*!\n\n# Great Job!", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "##### Copyright 2019 The TensorFlow Authors.", "_____no_output_____" ] ], [ [ "#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# https://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.", "_____no_output_____" ] ], [ [ "#Improving Computer Vision Accuracy using Convolutions\n\nIn the previous lessons you saw how to do fashion recognition using a Deep Neural Network (DNN) containing three layers -- the input layer (in the shape of the data), the output layer (in the shape of the desired output) and a hidden layer. You experimented with the impact of different sizes of hidden layer, number of training epochs etc on the final accuracy.\n\nFor convenience, here's the entire code again. Run it and take a note of the test accuracy that is printed out at the end. ", "_____no_output_____" ] ], [ [ "import tensorflow as tf\nmnist = tf.keras.datasets.fashion_mnist\n(training_images, training_labels), (test_images, test_labels) = mnist.load_data()\ntraining_images=training_images / 255.0\ntest_images=test_images / 255.0\nmodel = tf.keras.models.Sequential([\n tf.keras.layers.Flatten(),\n tf.keras.layers.Dense(128, activation=tf.nn.relu),\n tf.keras.layers.Dense(10, activation=tf.nn.softmax)\n])\nmodel.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\nmodel.fit(training_images, training_labels, epochs=5)\n\ntest_loss = model.evaluate(test_images, test_labels)", "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz\n32768/29515 [=================================] - 0s 0us/step\nDownloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz\n26427392/26421880 [==============================] - 0s 0us/step\nDownloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz\n8192/5148 [===============================================] - 0s 0us/step\nDownloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz\n4423680/4422102 [==============================] - 0s 0us/step\nEpoch 1/5\n1875/1875 [==============================] - 4s 2ms/step - loss: 0.4968 - accuracy: 0.8242\nEpoch 2/5\n1875/1875 [==============================] - 4s 2ms/step - loss: 0.3721 - accuracy: 0.8649\nEpoch 3/5\n1875/1875 [==============================] - 4s 2ms/step - loss: 0.3326 - accuracy: 0.8787\nEpoch 4/5\n1875/1875 [==============================] - 4s 2ms/step - loss: 0.3107 - accuracy: 0.8857\nEpoch 5/5\n1875/1875 [==============================] - 4s 2ms/step - loss: 0.2946 - accuracy: 0.8911\n313/313 [==============================] - 1s 1ms/step - loss: 0.3476 - accuracy: 0.8750\n" ] ], [ [ "Your accuracy is probably about 89% on training and 87% on validation...not bad...But how do you make that even better? One way is to use something called Convolutions. I'm not going to details on Convolutions here, but the ultimate concept is that they narrow down the content of the image to focus on specific, distinct, details. \n\nIf you've ever done image processing using a filter (like this: https://en.wikipedia.org/wiki/Kernel_(image_processing)) then convolutions will look very familiar.\n\nIn short, you take an array (usually 3x3 or 5x5) and pass it over the image. By changing the underlying pixels based on the formula within that matrix, you can do things like edge detection. So, for example, if you look at the above link, you'll see a 3x3 that is defined for edge detection where the middle cell is 8, and all of its neighbors are -1. In this case, for each pixel, you would multiply its value by 8, then subtract the value of each neighbor. Do this for every pixel, and you'll end up with a new image that has the edges enhanced.\n\nThis is perfect for computer vision, because often it's features that can get highlighted like this that distinguish one item for another, and the amount of information needed is then much less...because you'll just train on the highlighted features.\n\nThat's the concept of Convolutional Neural Networks. Add some layers to do convolution before you have the dense layers, and then the information going to the dense layers is more focussed, and possibly more accurate.\n\nRun the below code -- this is the same neural network as earlier, but this time with Convolutional layers added first. It will take longer, but look at the impact on the accuracy:", "_____no_output_____" ] ], [ [ "import tensorflow as tf\nprint(tf.__version__)\nmnist = tf.keras.datasets.fashion_mnist\n(training_images, training_labels), (test_images, test_labels) = mnist.load_data()\n# Reshaping the images to tell the convolutional layers that the images are in greyscale by adding an extra dimension of 1\ntraining_images=training_images.reshape(60000, 28, 28, 1)\ntraining_images=training_images / 255.0\ntest_images = test_images.reshape(10000, 28, 28, 1)\ntest_images=test_images/255.0\nmodel = tf.keras.models.Sequential([\n # Mind it, Convolutions and MaxPooling are always applied before the Deep Neural Network Layers\n # Why 2D? because, applied convolutions and maxpoolings are 2D array in nature (having rows and columns)\n # Here 64 is the total number of convolutional filters of size (3, 3) applied\n ### Be careful about the shapes!!! You obviously need to mention the input_shape at the first Conv2D(), otherwise it will turn\n ### into an error!!! \n tf.keras.layers.Conv2D(64, (3, 3), activation='relu', input_shape=(28, 28, 1)),\n tf.keras.layers.MaxPooling2D(2, 2),\n tf.keras.layers.Conv2D(64, (3, 3), activation='relu'),\n tf.keras.layers.MaxPooling2D(2, 2),\n tf.keras.layers.Flatten(),\n tf.keras.layers.Dense(128, activation='relu'),\n tf.keras.layers.Dense(10, activation='softmax')\n])\nmodel.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n# Mind it, model.summary() is a way of cross-verification that the DNN with Convolutions is applied correctly with accurate\n# shape retention\nmodel.summary()\nmodel.fit(training_images, training_labels, epochs=5)\ntest_loss = model.evaluate(test_images, test_labels)\n", "2.5.0\nModel: \"sequential_3\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\nconv2d_4 (Conv2D) (None, 26, 26, 64) 640 \n_________________________________________________________________\nmax_pooling2d_4 (MaxPooling2 (None, 13, 13, 64) 0 \n_________________________________________________________________\nconv2d_5 (Conv2D) (None, 11, 11, 64) 36928 \n_________________________________________________________________\nmax_pooling2d_5 (MaxPooling2 (None, 5, 5, 64) 0 \n_________________________________________________________________\nflatten_3 (Flatten) (None, 1600) 0 \n_________________________________________________________________\ndense_6 (Dense) (None, 128) 204928 \n_________________________________________________________________\ndense_7 (Dense) (None, 10) 1290 \n=================================================================\nTotal params: 243,786\nTrainable params: 243,786\nNon-trainable params: 0\n_________________________________________________________________\nEpoch 1/5\n1875/1875 [==============================] - 81s 43ms/step - loss: 0.4397 - accuracy: 0.8383\nEpoch 2/5\n1875/1875 [==============================] - 79s 42ms/step - loss: 0.2954 - accuracy: 0.8918\nEpoch 3/5\n1875/1875 [==============================] - 79s 42ms/step - loss: 0.2518 - accuracy: 0.9076\nEpoch 4/5\n1875/1875 [==============================] - 78s 42ms/step - loss: 0.2214 - accuracy: 0.9164\nEpoch 5/5\n1875/1875 [==============================] - 78s 42ms/step - loss: 0.1924 - accuracy: 0.9273\n313/313 [==============================] - 4s 12ms/step - loss: 0.2676 - accuracy: 0.9057\n" ] ], [ [ "It's likely gone up to about 93% on the training data and 91% on the validation data. \n\nThat's significant, and a step in the right direction!\n\nTry running it for more epochs -- say about 20, and explore the results! But while the results might seem really good, the validation results may actually go down, due to something called 'overfitting' which will be discussed later. \n\n(In a nutshell, 'overfitting' occurs when the network learns the data from the training set really well, but it's too specialised to only that data, and as a result is less effective at seeing *other* data. For example, if all your life you only saw red shoes, then when you see a red shoe you would be very good at identifying it, but blue suade shoes might confuse you...and you know you should never mess with my blue suede shoes.)\n\nThen, look at the code again, and see, step by step how the Convolutions were built:", "_____no_output_____" ], [ "Step 1 is to gather the data. You'll notice that there's a bit of a change here in that the training data needed to be reshaped. That's because the first convolution expects a single tensor containing everything, so instead of 60,000 28x28x1 items in a list, we have a single 4D list that is 60,000x28x28x1, and the same for the test images. If you don't do this, you'll get an error when training as the Convolutions do not recognize the shape. \n\n\n\n```\nimport tensorflow as tf\nmnist = tf.keras.datasets.fashion_mnist\n(training_images, training_labels), (test_images, test_labels) = mnist.load_data()\ntraining_images=training_images.reshape(60000, 28, 28, 1)\ntraining_images=training_images / 255.0\ntest_images = test_images.reshape(10000, 28, 28, 1)\ntest_images=test_images/255.0\n```\n", "_____no_output_____" ], [ "Next is to define your model. Now instead of the input layer at the top, you're going to add a Convolution. The parameters are:\n\n1. The number of convolutions you want to generate. Purely arbitrary, but good to start with something in the order of 32\n2. The size of the Convolution, in this case a 3x3 grid\n3. The activation function to use -- in this case we'll use relu, which you might recall is the equivalent of returning x when x>0, else returning 0\n4. In the first layer, the shape of the input data.\n\nYou'll follow the Convolution with a MaxPooling layer which is then designed to compress the image, while maintaining the content of the features that were highlighted by the convlution. By specifying (2,2) for the MaxPooling, the effect is to quarter the size of the image. Without going into too much detail here, the idea is that it creates a 2x2 array of pixels, and picks the biggest one, thus turning 4 pixels into 1. It repeats this across the image, and in so doing halves the number of horizontal, and halves the number of vertical pixels, effectively reducing the image by 25%.\n\nYou can call model.summary() to see the size and shape of the network, and you'll notice that after every MaxPooling layer, the image size is reduced in this way. \n\n\n```\nmodel = tf.keras.models.Sequential([\n tf.keras.layers.Conv2D(32, (3,3), activation='relu', input_shape=(28, 28, 1)),\n tf.keras.layers.MaxPooling2D(2, 2),\n```\n", "_____no_output_____" ], [ "Add another convolution\n\n\n\n```\n tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n tf.keras.layers.MaxPooling2D(2,2)\n```\n", "_____no_output_____" ], [ "Now flatten the output. After this you'll just have the same DNN structure as the non convolutional version\n\n```\n tf.keras.layers.Flatten(),\n```\n", "_____no_output_____" ], [ "The same 128 dense layers, and 10 output layers as in the pre-convolution example:\n\n\n\n```\n tf.keras.layers.Dense(128, activation='relu'),\n tf.keras.layers.Dense(10, activation='softmax')\n])\n```\n", "_____no_output_____" ], [ "Now compile the model, call the fit method to do the training, and evaluate the loss and accuracy from the test set.\n\n\n\n```\nmodel.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\nmodel.fit(training_images, training_labels, epochs=5)\ntest_loss, test_acc = model.evaluate(test_images, test_labels)\nprint(test_acc)\n```\n", "_____no_output_____" ], [ "# Visualizing the Convolutions and Pooling\n\nThis code will show us the convolutions graphically. The print (test_labels[;100]) shows us the first 100 labels in the test set, and you can see that the ones at index 0, index 23 and index 28 are all the same value (9). They're all shoes. Let's take a look at the result of running the convolution on each, and you'll begin to see common features between them emerge. Now, when the DNN is training on that data, it's working with a lot less, and it's perhaps finding a commonality between shoes based on this convolution/pooling combination.", "_____no_output_____" ] ], [ [ "print(test_labels[:100])", "[9 2 1 1 6 1 4 6 5 7 4 5 7 3 4 1 2 4 8 0 2 5 7 9 1 4 6 0 9 3 8 8 3 3 8 0 7\n 5 7 9 6 1 3 7 6 7 2 1 2 2 4 4 5 8 2 2 8 4 8 0 7 7 8 5 1 1 2 3 9 8 7 0 2 6\n 2 3 1 2 8 4 1 8 5 9 5 0 3 2 0 6 5 3 6 7 1 8 0 1 4 2]\n" ], [ "import matplotlib.pyplot as plt\nf, axarr = plt.subplots(3,4)\nFIRST_IMAGE=0\nSECOND_IMAGE=7\nTHIRD_IMAGE=26\nCONVOLUTION_NUMBER = 1\nfrom tensorflow.keras import models\nlayer_outputs = [layer.output for layer in model.layers]\nactivation_model = tf.keras.models.Model(inputs = model.input, outputs = layer_outputs)\nfor x in range(0,4):\n f1 = activation_model.predict(test_images[FIRST_IMAGE].reshape(1, 28, 28, 1))[x]\n axarr[0,x].imshow(f1[0, : , :, CONVOLUTION_NUMBER], cmap='inferno')\n axarr[0,x].grid(False)\n f2 = activation_model.predict(test_images[SECOND_IMAGE].reshape(1, 28, 28, 1))[x]\n axarr[1,x].imshow(f2[0, : , :, CONVOLUTION_NUMBER], cmap='inferno')\n axarr[1,x].grid(False)\n f3 = activation_model.predict(test_images[THIRD_IMAGE].reshape(1, 28, 28, 1))[x]\n axarr[2,x].imshow(f3[0, : , :, CONVOLUTION_NUMBER], cmap='inferno')\n axarr[2,x].grid(False)", "_____no_output_____" ] ], [ [ "EXERCISES\n\n1. Try editing the convolutions. Change the 32s to either 16 or 64. What impact will this have on accuracy and/or training time.\n\n2. Remove the final Convolution. What impact will this have on accuracy or training time?\n\n3. How about adding more Convolutions? What impact do you think this will have? Experiment with it.\n\n4. Remove all Convolutions but the first. What impact do you think this will have? Experiment with it. \n\n5. In the previous lesson you implemented a callback to check on the loss function and to cancel training once it hit a certain amount. See if you can implement that here!", "_____no_output_____" ] ], [ [ "import tensorflow as tf\nprint(tf.__version__)\nmnist = tf.keras.datasets.mnist\n(training_images, training_labels), (test_images, test_labels) = mnist.load_data()\ntraining_images=training_images.reshape(60000, 28, 28, 1)\ntraining_images=training_images / 255.0\ntest_images = test_images.reshape(10000, 28, 28, 1)\ntest_images=test_images/255.0\nmodel = tf.keras.models.Sequential([\n tf.keras.layers.Conv2D(32, (3,3), activation='relu', input_shape=(28, 28, 1)),\n tf.keras.layers.MaxPooling2D(2, 2),\n tf.keras.layers.Flatten(),\n tf.keras.layers.Dense(128, activation='relu'),\n tf.keras.layers.Dense(10, activation='softmax')\n])\nmodel.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\nmodel.fit(training_images, training_labels, epochs=10)\ntest_loss, test_acc = model.evaluate(test_images, test_labels)\nprint(test_acc)", "2.5.0\nDownloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n11493376/11490434 [==============================] - 0s 0us/step\nEpoch 1/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.1424 - accuracy: 0.9569\nEpoch 2/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0502 - accuracy: 0.9842\nEpoch 3/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0311 - accuracy: 0.9901\nEpoch 4/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0217 - accuracy: 0.9932\nEpoch 5/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0138 - accuracy: 0.9955\nEpoch 6/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0100 - accuracy: 0.9969\nEpoch 7/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0078 - accuracy: 0.9974\nEpoch 8/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0063 - accuracy: 0.9980\nEpoch 9/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0046 - accuracy: 0.9986\nEpoch 10/10\n1875/1875 [==============================] - 34s 18ms/step - loss: 0.0055 - accuracy: 0.9982\n313/313 [==============================] - 2s 6ms/step - loss: 0.0584 - accuracy: 0.9866\n0.9865999817848206\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Fundamentals, introduction to machine learning\n\n\nThe purpose of these guides is to go a bit deeper into the details behind common machine learning methods, assuming little math background, and teach you how to use popular machine learning Python packages. In particular, we'll focus on the Numpy and PyTorch libraries.\n\nI'll assume you have some experience programming with Python -- if not, check out the initial [fundamentals of Python guide](https://github.com/ml4a/ml4a-guides/blob/master/notebooks/intro_python.ipynb) or for a longer, more comprehensive resource: [Learn Python the Hard Way](http://learnpythonthehardway.org/book/). It will really help to illustrate the concepts introduced here.\n\nNumpy underlies most Python machine learning packages and is great for performing quick sketches or working through calculations. PyTorch rivals alternative libraries, such as TensorFlow, for its flexibility and ease of use. Despite the high level appearance of PyTorch, it can be quite low-level, which is great for experimenting with novel algorithms. PyTorch can seamlessly be integrated with distributed computation libraries, like Ray, to make the Kessel Run in less than 12 parsecs (citation needed). \n\nThese guides will present the formal math for concepts alongside Python code examples since this often (for me at least) is a lot easier to develop an intuition for. Each guide is also available as an iPython notebook for your own experimentation.\n\nThe guides are not meant to exhaustively cover the field of machine learning but I hope they will instill you with the confidence and knowledge to explore further on your own.\n\nIf you do want more details, you might enjoy my [artificial intelligence notes](http://frnsys.com/ai_notes).\n\n\n### Modeling the world\n\nYou've probably seen various machine learning algorithms pop up -- linear regression, SVMs, neural networks, random forests, etc. How are they all related? What do they have in common? What is machine learning for anyways?\n\nFirst, let's consider the general, fundamental problem all machine learning is concerned with, leaving aside the algorithm name soup for now. The primary concern of machine learning is _modeling the world_.\n\nWe can model phenomena or systems -- both natural and artificial, if you want to make that distinction -- with mathematical functions. We see something out in the world and want to describe it in some way, we want to formalize how two or more things are related, and we can do that with a function. The problem is, for a given phenomenon, how do we figure out what function to use? There are infinitely many to choose from!\n\nBefore this gets too abstract, let's use an example to make things more concrete.\n\nSay we have a bunch of data about the heights and weights of a species of deer. We want to understand how these two variables are related -- in particular, given the weight of a deer, can we predict its height?\n\nYou might see where this is going. The data looks like a line, and lines in general are described by functions of the form $y = mx + b$.\n\nRemember that lines vary depending on what the values of $m$ and $b$ are:\n\n\n\nThus $m$ and $b$ uniquely define a function -- thus they are called the _parameters_ of the function -- and when it comes to machine learning, these parameters are what we ultimately want to learn. So when I say there are infinitely many functions to choose from, it is because $m$ and $b$ can pretty much take on any value. Machine learning techniques essentially search through these possible functions to find parameters that best fit the data you have. One way machine learning algorithms are differentiated is by how exactly they conduct this search (i.e. how they learn parameters).\n\nIn this case we've (reasonably) assumed the function takes the form $y = mx + b$, but conceivably you may have data that doesn't take the form of a line. Real world data is typically a lot more convoluted-looking. Maybe the true function has a $sin$ in it, for example.\n\nThis is where another main distinction between machine learning algorithms comes in -- certain algorithms can model only certain forms of functions. _Linear regression_, for example, can only model linear functions, as indicated by its name. Neural networks, on the other hand, are _universal function approximators_, which mean they can (in theory) approximate _any_ function, no matter how exotic. This doesn't necessarily make them a better method, just better suited for certain circumstances (there are many other considerations when choosing an algorithm).\n\nFor now, let's return to the line function. Now that we've looked at the $m$ and $b$ variables, let's consider the input variable $x$. A function takes a numerical input; that is $x$ must be a number of some kind. That's pretty straightforward here since the deer weights are already numbers. But this is not always the case! What if we want to predict the sales price of a house. A house is not a number. We have to find a way to _represent_ it as a number (or as several numbers, i.e. a vector, which will be detailed in a moment), e.g. by its square footage. This challenge of representation is a major part of machine learning; the practice of building representations is known as _feature engineering_ since each variable (e.g. square footage or zip code) used for the representation is called a _feature_.\n\nIf you think about it, representation is a practice we regularly engage in. The word \"house\" is not a house any more than an image of a house is -- there is no true \"house\" anyways, it is always a constellation of various physical and nonphysical components.\n\nThat's about it -- broadly speaking, machine learning is basically a bunch of algorithms that learn you a function, which is to say they learn the parameters that uniquely define a function.\n\n### Vectors\n\nIn the line example before I mentioned that we might have multiple numbers representing an input. For example, a house probably can't be solely represented by its square footage -- perhaps we also want to consider how many bedrooms it has, or how high the ceilings are, or its distance from local transportation. How do we group these numbers together?\n\nThat's what _vectors_ are for (they come up for many other reasons too, but we'll focus on representation for now). Vectors, along with matrices and other tensors (which will be explained a bit further down), could be considered the \"primitives\" of machine learning.\n\nThe Numpy library is best for dealing with vectors (and other tensors) in Python. A more complete introduction to Numpy is provided in the [numpy and basic mathematics guide](https://github.com/ml4a/ml4a-guides/blob/master/notebooks/math_review_numpy.ipynb).\n\nLet's import `numpy` with the alias `nf`:", "_____no_output_____" ] ], [ [ "import numpy as np", "_____no_output_____" ] ], [ [ "You may have encountered vectors before in high school or college -- to use Python terms, a vector is like a list of numbers. The mathematical notation is quite similar to Python code, e.g. `[5,4]`, but `numpy` has its own way of instantiating a vector:", "_____no_output_____" ] ], [ [ "v = np.array([5, 4])", "_____no_output_____" ] ], [ [ "$$\nv = \\begin{bmatrix} 5 \\\\ 4 \\end{bmatrix}\n$$\n\nVectors are usually represented with lowercase variables.\n\nNote that we never specified how _many_ numbers (also called _components_) a vector has - because it can have any amount. The amount of components a vector has is called its _dimensionality_. The example vector above has two dimensions. The vector `x = [8,1,3]` has three dimensions, and so on. Components are usually indicated by their index (usually using 1-indexing), e.g. in the previous vector, $x_1$ refers to the value $8$.\n\n\"Dimensions\" in the context of vectors is just like the spatial dimensions you spend every day in. These dimensions define a __space__, so a two-dimensional vector, e.g. `[5,4]`, can describe a point in 2D space and a three-dimensional vector, e.g. `[8,1,3]`, can describe a point in 3D space. As mentioned before, there is no limit to the amount of dimensions a vector may have (technically, there must be one or more dimensions), so we could conceivably have space consisting of thousands or tens of thousands of dimensions. At that point we can't rely on the same human intuitions about space as we could when working with just two or three dimensions. In practice, most interesting applications of machine learning deal with many, many dimensions.\n\nWe can get a better sense of this by plotting a vector out. For instance, a 2D vector `[5,0]` would look like:\n\n\n\nSo in a sense vectors can be thought of lines that \"point\" to the position they specify - here the vector is a line \"pointing\" to `[5,0]`. If the vector were 3D, e.g. `[8,1,3]`, then we would have to visualize it in 3D space, and so on.\n\nSo vectors are great - they allow us to form logical groupings of numbers. For instance, if we're talking about cities on a map we would want to group their latitude and longitude together. We'd represent Lagos with `[6.455027, 3.384082]` and Beijing separately with `[39.9042, 116.4074]`. If we have an inventory of books for sale, we could represent each book with its own vector consisting of its price, number of pages, and remaining stock.\n\nTo use vectors in functions, there are a few mathematical operations you need to know.\n\n### Basic vector operations\n\nVectors can be added (and subtracted) easily:", "_____no_output_____" ] ], [ [ "np.array([6, 2]) + np.array([-4, 4])", "_____no_output_____" ] ], [ [ "$$\n\\begin{bmatrix} 6 \\\\ 2 \\end{bmatrix} + \\begin{bmatrix} -4 \\\\ 4 \\end{bmatrix} = \\begin{bmatrix} 6 + -4 \\\\ 2 + 4 \\end{bmatrix} = \\begin{bmatrix} 2 \\\\ 6 \\end{bmatrix}\n$$\n\nHowever, when it comes to vector multiplication there are many different kinds.\n\nThe simplest is _vector-scalar_ multiplication:", "_____no_output_____" ] ], [ [ "3 * np.array([2, 1])", "_____no_output_____" ] ], [ [ "$$\n3\\begin{bmatrix} 2 \\\\ 1 \\end{bmatrix} = \\begin{bmatrix} 3 \\times 2 \\\\ 3 \\times 1\n\\end{bmatrix} = \\begin{bmatrix} 6 \\\\ 3 \\end{bmatrix}\n$$\n\nBut when you multiply two vectors together you have a few options. I'll cover the two most important ones here.\n\nThe one you might have thought of is the _element-wise product_, also called the _pointwise product_, _component-wise product_, or the _Hadamard product_, typically notated with $\\odot$. This just involves multiplying the corresponding elements of each vector together, resulting in another vector:", "_____no_output_____" ] ], [ [ "np.array([6, 2]) * np.array([-4, 4])", "_____no_output_____" ] ], [ [ "$$\n\\begin{bmatrix} 6 \\\\ 2 \\end{bmatrix} \\odot \\begin{bmatrix} -4 \\\\ 4 \\end{bmatrix} = \\begin{bmatrix} 6 \\times -4 \\\\ 2 \\times 4 \\end{bmatrix} = \\begin{bmatrix} -24 \\\\ 8 \\end{bmatrix}\n$$\n\nThe other vector product, which you'll encounter a lot, is the _dot product_, also called _inner product_, usually notated with $\\cdot$ (though when vectors are placed side-by-side this often implies dot multiplication). This involves multiplying corresponding elements of each vector and then summing the resulting vector's components (so this results in a scalar rather than another vector).", "_____no_output_____" ] ], [ [ "np.dot(np.array([6, 2]), np.array([-4, 4]))", "_____no_output_____" ] ], [ [ "$$\n\\begin{bmatrix} 6 \\\\ 2 \\end{bmatrix} \\cdot \\begin{bmatrix} -4 \\\\ 4 \\end{bmatrix} = (6 \\times -4) + (2 \\times 4) = -16\n$$\n\nThe more general formulation is:", "_____no_output_____" ] ], [ [ "# a slow pure-Python dot product\ndef dot(a, b):\n assert len(a) == len(b)\n return sum(a_i * b_i for a_i, b_i in zip(a,b))", "_____no_output_____" ] ], [ [ "$$\n\\begin{aligned}\n\\vec{a} \\cdot \\vec{b} &= \\begin{bmatrix} a_1 \\\\ a_2 \\\\ \\vdots \\\\ a_n \\end{bmatrix} \\cdot \\begin{bmatrix} b_1 \\\\ b_2 \\\\ \\vdots \\\\ b_n \\end{bmatrix} = a_1b_1 + a_2b_2 + \\dots + a_nb_n \\\\\n&= \\sum^n_{i=1} a_i b_i\n\\end{aligned}\n$$\n\nNote that the vectors in these operations must have the same dimensions!\n\nPerhaps the most important vector operation mentioned here is the dot product. We'll return to the house example to see why. Let's say want to represent a house with three variables: square footage, number of bedrooms, and the number of bathrooms. For convenience we'll notate the variables $x_1, x_2, x_3$, respectively. We're working in three dimensions now so instead of learning a line we're learning a _hyperplane_ (if we were working with two dimensions we'd be learning a plane, \"hyperplane\" is the term for the equivalent of a plane in higher dimensions).\n\nAside from the different name, the function we're learning is essentially of the same form as before, just with more variables and thus more parameters. We'll notate each parameter as $\\theta_i$ as is the convention (you may see $\\beta_i$ used elsewhere), and for the intercept (what was the $b$ term in the original line), we'll add in a dummy variable $x_0 = 1$ as is the typical practice (thus $\\theta_0$ is equivalent to $b$):", "_____no_output_____" ] ], [ [ "# this is so clumsy in python;\n# this will become more concise in a bit\ndef f(x0, x1, x2, x3, theta0, theta1, theta2, theta3):\n return theta0 * x0\\\n + theta1 * x1\\\n + theta2 * x2\\\n + theta3 * x3", "_____no_output_____" ] ], [ [ "$$\ny = \\theta_0 x_0 + \\theta_1 x_1 + \\theta_2 x_2 + \\theta_3 x_3\n$$\n\nThis kind of looks like the dot product, doesn't it? In fact, we can re-write this entire function as a dot product. We define our feature vector $x = [x_0, x_1, x_2, x_3]$ and our parameter vector $\\theta = [\\theta_0, \\theta_1, \\theta_2, \\theta_3]$, then re-write the function:", "_____no_output_____" ] ], [ [ "def f(x, theta):\n return x.dot(theta)", "_____no_output_____" ] ], [ [ "$$\ny = \\theta x\n$$\n\nSo that's how we incorporate multiple features in a representation.\n\nThere's a whole lot more to vectors than what's presented here, but this is the ground-level knowledge you should have of them. Other aspects of vectors will be explained as they come up.\n\n## Learning\n\nSo machine learning algorithms learn parameters - how do they do it?\n\nHere we're focusing on the most common kind of machine learning - _supervised_ learning. In supervised learning, the algorithm learns parameters from data which includes both the inputs and the true outputs. This data is called _training_ data.\n\nAlthough they vary on specifics, there is a general approach that supervised machine learning algorithms use to learn parameters. The idea is that the algorithm takes an input example, inputs it into the current guess at the function (called the _hypothesis_, notate $h_{\\theta}$), and then checks how wrong its output is against the true output. The algorithm then updates its hypothesis (that is, its guesses for the parameters), accordingly.\n\n\"How wrong\" an algorithm is, can vary depending on the _loss function_ it is using. The loss function takes the algorithm's current guess for the output, $\\hat y$, and the true output, $y$, and returns some value quantifying its wrongness. Certain loss functions are more appropriate for certain tasks, which we'll get into later.\n\nWe'll get into the specifies of how the algorithm determines what kind of update to perform (i.e. how much each parameter changes), but before we do that we should consider how we manage batches of training examples (i.e. multiple training vectors) simultaneously.\n\n\n## Matrices\n\n__Matrices__ are in a sense a \"vector\" of vectors. That is, where a vector can be thought of as a logical grouping of numbers, a matrix can be thought of as a logical grouping of vectors. So if a vector represents a book in our catalog (id, price, number in stock), a matrix could represent the entire catalog (each row refers to a book). Or if we want to represent a grayscale image, the matrix can represent the brightness values of the pixels in the image.", "_____no_output_____" ] ], [ [ "A = np.array([\n [6, 8, 0],\n [8, 2, 7],\n [3, 3, 9],\n [3, 8, 6]\n])", "_____no_output_____" ] ], [ [ "$$\n\\mathbf A =\n\\begin{bmatrix}\n6 & 8 & 0 \\\\\n8 & 2 & 7 \\\\\n3 & 3 & 9 \\\\\n3 & 8 & 6\n\\end{bmatrix}\n$$\n\nMatrices are usually represented with uppercase variables.\n\nNote that the \"vectors\" in the matrix must have the same dimension. The matrix's dimensions are expressed in the form $m \\times n$, meaning that there are $m$ rows and $n$ columns. So the example matrix has dimensions of $4 \\times 3$. Numpy calls these dimensions a matrix's \"shape\".\n\nWe can access a particular element, $A_{i,j}$, in a matrix by its indices. Say we want to refer to the element in the 2nd row and the 3rd column (remember that python uses 0-indexing):", "_____no_output_____" ] ], [ [ "A[1,2]", "_____no_output_____" ] ], [ [ "### Basic matrix operations\n\nLike vectors, matrix addition and subtraction is straightforward (again, they must be of the same dimensions):", "_____no_output_____" ] ], [ [ "B = np.array([\n [8, 3, 7],\n [2, 9, 6],\n [2, 5, 6],\n [5, 0, 6]\n])\n\nA + B", "_____no_output_____" ] ], [ [ "$$\n\\begin{aligned}\n\\mathbf B &=\n\\begin{bmatrix}\n8 & 3 & 7 \\\\\n2 & 9 & 6 \\\\\n2 & 5 & 6 \\\\\n5 & 0 & 6\n\\end{bmatrix} \\\\\nA + B &=\n\\begin{bmatrix}\n8+6 & 3+8 & 7+0 \\\\\n2+8 & 9+2 & 6+7 \\\\\n2+3 & 5+3 & 6+9 \\\\\n5+3 & 0+8 & 6+6\n\\end{bmatrix} \\\\\n&=\n\\begin{bmatrix}\n14 & 11 & 7 \\\\\n10 & 11 & 13 \\\\\n5 & 8 & 15 \\\\\n8 & 8 & 12\n\\end{bmatrix} \\\\\n\\end{aligned}\n$$\n\nMatrices also have a few different multiplication operations, like vectors.\n\n_Matrix-scalar multiplication_ is similar to vector-scalar multiplication - you just distribute the scalar, multiplying it with each element in the matrix.\n\n_Matrix-vector products_ require that the vector has the same dimension as the matrix has columns, i.e. for an $m \\times n$ matrix, the vector must be $n$-dimensional. The operation basically involves taking the dot product of each matrix row with the vector:", "_____no_output_____" ] ], [ [ "# a slow pure-Python matrix-vector product,\n# using our previous dot product implementation\ndef matrix_vector_product(M, v):\n return [np.dot(row, v) for row in M]\n\n# or, with numpy, you could use np.matmul(A,v)", "_____no_output_____" ] ], [ [ "$$\n\\mathbf M v =\n\\begin{bmatrix}\nM_{1} \\cdot v \\\\\n\\vdots \\\\\nM_{m} \\cdot v \\\\\n\\end{bmatrix}\n$$\n\nWe have a few options when it comes to multiplying matrices with matrices.\n\nHowever, before we go any further we should talk about the _tranpose_ operation - this just involves switching the columns and rows of a matrix. The transpose of a matrix $A$ is notated $A^T$:", "_____no_output_____" ] ], [ [ "A = np.array([\n [1,2,3],\n [4,5,6]\n ])\n\nnp.transpose(A)", "_____no_output_____" ] ], [ [ "$$\n\\begin{aligned}\n\\mathbf A &=\n\\begin{bmatrix}\n1 & 2 & 3 \\\\\n4 & 5 & 6\n\\end{bmatrix} \\\\\n\\mathbf A^T &=\n\\begin{bmatrix}\n1 & 4 \\\\\n2 & 5 \\\\\n3 & 6\n\\end{bmatrix}\n\\end{aligned}\n$$\n\n\nFor matrix-matrix products, the matrix on the lefthand must have the same number of columns as the righthand's rows. To be more concrete, we'll represent a matrix-matrix product as $A B$ and we'll say that $A$ has $m \\times n$ dimensions. For this operation to work, $B$ must have $n \\times p$ dimensions. The resulting product will have $m \\times p$ dimensions.", "_____no_output_____" ] ], [ [ "# a slow pure-Python matrix Hadamard product\ndef matrix_matrix_product(A, B):\n _, a_cols = np.shape(A)\n b_rows, _ = np.shape(B)\n assert a_cols == b_rows\n\n result = []\n # tranpose B so we can iterate over its columns\n for col in np.tranpose(B):\n # using our previous implementation\n result.append(\n matrix_vector_product(A, col))\n return np.transpose(result)", "_____no_output_____" ] ], [ [ "$$\n\\mathbf AB =\n\\begin{bmatrix}\nA B^T_1 \\\\\n\\vdots \\\\\nA B^T_p\n\\end{bmatrix}^T\n$$\n\nFinally, like with vectors, we also have Hadamard (element-wise) products:", "_____no_output_____" ] ], [ [ "# a slow pure-Python matrix-matrix product\n# or, with numpy, you can use A * B\ndef matrix_matrix_hadamard(A, B):\n result = []\n for a_row, b_row in zip(A, B):\n result.append(\n zip(a_i * b_i for a_i, b_i in zip(a_row, b_row)))", "_____no_output_____" ] ], [ [ "$$\n\\mathbf A \\odot B =\n\\begin{bmatrix}\nA_{1,1} B_{1,1} & \\dots & A_{1,n} B_{1,n} \\\\\n\\vdots & \\dots & \\vdots \\\\\nA_{m,1} B_{m,1} & \\dots & A_{m,n} B_{m,n}\n\\end{bmatrix}\n$$\n\nLike vector Hadamard products, this requires that the two matrices share the same dimensions.\n\n## Tensors\n\nWe've seen vectors, which is like a list of numbers, and matrices, which is like a list of a list of numbers. We can generalize this concept even further, for instance, with a list of a list of a list of numbers and so on. What all of these structures are called are _tensors_ (i.e. the \"tensor\" in \"TensorFlow\"). They are distinguished by their _rank_, which, if you're thinking in the \"list of lists\" way, refers to the number of nestings. So a vector has a rank of one (just a list of numbers) and a matrix has a rank of two (a list of a list of numbers).\n\nAnother way to think of rank is by number of indices necessary to access an element in the tensor. An element in a vector is accessed by one index, e.g. `v[i]`, so it is of rank one. An element in a matrix is accessed by two indices, e.g. `M[i,j]`, so it is of rank two.\n\nWhy is the concept of a tensor useful? Before we referred to vectors as a logical grouping of numbers and matrices as a logical grouping of vectors. What if we need a logical grouping of matrices? That's what 3rd-rank tensors are! A matrix can represent a grayscale image, but what about a color image with three color channels (red, green, blue)? With a 3rd-rank tensor, we could represent each channel as its own matrix and group them together.\n\n## Learning continued\n\nWhen the current hypothesis is wrong, how does the algorithm know how to adjust the parameters?\n\nLet's take a step back and look at it another way. The loss function measures the wrongness of the hypothesis $h_{\\theta}$ - another way of saying this is the loss function is a function of the parameters $\\theta$. So we could notate it as $L(\\theta)$.\n\nThe minimum of $L(\\theta)$ is the point where the parameters guess $\\theta$ is least wrong (at best, $L(\\theta) = 0$, i.e. a perfect score, though this is not always good, as will be explained later); i.e. the best guess for the parameters.\n\nSo the algorithm learns the best-fitting function by minimizing its loss function. That is, we can frame this as an optimization problem.\n\nThere are many techniques to solve an optimization problem - sometimes they can be solved analytically (i.e. by moving around variables and isolating the one you want to solve for), but more often than not we must solve them numerically, i.e. by guessing a lot of different values - but not randomly!\n\nThe prevailing technique now is called _gradient descent_, and to understand how it works, we have to understand derivatives.\n\n## Derivatives\n\nDerivatives are everywhere in machine learning, so it's worthwhile become a bit familiar with them. I won't go into specifics on differentiation (how to calculate derivatives) because now we're spoiled with automatic differentiation, but it's still good to have a solid intuition about derivatives themselves.\n\nA derivative expresses a rate of (instantaneous) change - they are always about how one variable quantity changes with respect to another variable quantity. That's basically all there is to it. For instance, velocity is a derivative which expresses how position changes with respect to time. Another interpretation, which is more relevant to machine learning, is that a derivative tells us how to change one variable to achieve a desired change in the other variable. Velocity, for instance, tells us how to change position by \"changing\" time.\n\nTo get a better understanding of _instantaneous_ change, consider a cyclist, cycling on a line. We have data about their position over time. We could calculate an average velocity over the data's entire time period, but we typically prefer to know the velocity at any given _moment_ (i.e. at any _instant_).\n\nLet's get more concrete first. Let's say we have data for $n$ seconds, i.e. from $t_0$ to $t_n$ seconds, and the position at any given second $i$ is $p_i$. If we wanted to get the rate of change in position over the entire time interval, we'd just do:", "_____no_output_____" ] ], [ [ "positions = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, # moving forward\n 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, # pausing\n 9, 8, 7, 6, 5, 4, 3, 2, 1, 0] # moving backwards\nt_0 = 0\nt_n = 29\n(positions[t_n] - positions[t_0])/t_n", "_____no_output_____" ] ], [ [ "$$\nv = \\frac{p_n - p_0}{n}\n$$\n\nThis kind of makes it look like the cyclist didn't move at all. It would probably be more useful to identify the velocity at a given second $t$. Thus we want to come up with some function $v(t)$ which gives us the velocity at some second $t$. We can apply the same approach we just used to get the velocity over the entire time interval, but we focus on a shorter time interval instead. To get the _instantaneous_ change at $t$ we just keep reducing the interval we look at until it is basically 0.\n\nDerivatives have a special notation. A derivative of a function $f(x)$ with respect to a variable $x$ is notated:\n\n$$\n\\frac{\\delta f(x)}{\\delta x}\n$$\n\nSo if position is a function of time, e.g. $p = f(t)$, then velocity can be represented as $\\frac{\\delta p}{\\delta t}$. To drive the point home, this derivative is also a function of time (derivatives are functions of what their \"with respect to\" variable is).\n\nSince we are often computing derivatives of a function with respect to its input, a shorthand for the derivative of a function $f(x)$ with respect to $x$ can also be notated $f'(x)$.\n\n### The Chain Rule\n\nA very important property of derivatives is the _chain rule_ (there are other \"chain rules\" throughout mathematics, if we want to be specific, this is the \"chain rule of derivatives\"). The chain rule is important because it allows us to take complicated nested functions and more manageably differentiate them.\n\nLet's look at an example to make this concrete:", "_____no_output_____" ] ], [ [ "def g(x):\n return x**2\n\ndef h(x):\n return x**3\n\ndef f(x):\n return g(h(x))\n\n# derivatives\ndef g_(x):\n return 2*x\n\ndef h_(x):\n return 3*(x**2)", "_____no_output_____" ] ], [ [ "$$\n\\begin{aligned}\ng(x) &= x^2 \\\\\nh(x) &= x^3 \\\\\nf(x) &= g(h(x)) \\\\\ng'(x) &= 2x \\\\\nh'(x) &= 3x^2\n\\end{aligned}\n$$\n\nWe're interested in understanding how $f(x)$ changes with respect to $x$, so we want to compute the derivative of $f(x)$. The chain rule allows us to individually differentiate the component functions of $f(x)$ and multiply those to get $f'(x)$:", "_____no_output_____" ] ], [ [ "def f_(x):\n return g_(x) * h_(x)", "_____no_output_____" ] ], [ [ "$$\n\\frac{df}{dx} = \\frac{dg}{dh} \\frac{dh}{dx}\n$$\n\nThis example is a bit contrived (there is a very easy way to differentiate this particular example that doesn't involve the chain rule) but if $g(x)$ and $h(x)$ were really nasty functions, the chain rule makes them quite a lot easier to deal with.\n\nThe chain rule can be applied to nested functions ad nauseaum! You can apply it to something crazy like $f(g(h(u(q(p(x))))))$. In fact, with deep neural networks, you are typically dealing with function compositions even more gnarly than this, so the chain rule is cornerstone there.\n\n\n### Partial derivatives and gradients\n\nThe functions we've looked at so far just have a single input, but you can imagine many scenarios where you'd want to work with functions with some arbitrary number of inputs (i.e. a _multivariable_ function), like $f(x,y,z)$.\n\nHere's where _partial deriatives_ come into play. Partial derivatives are just like regular derivatives except we use them for multivariable functions; it just means we only differentiate with respect to one variable at a time. So for $f(x,y,z)$, we'd have a partial derivative with respect to $x$, i.e. $\\frac{\\partial f}{\\partial x}$ (note the slightly different notation), one with respect to $y$, i.e. $\\frac{\\partial f}{\\partial y}$, and one with respect to $z$, i.e. $\\frac{\\partial f}{\\partial z}$.\n\nThat's pretty simple! But it would be useful to group these partial derivatives together in some way. If we put these partial derivatives together in a vector, the resulting vector is the _gradient_ of $f$, notated $\\nabla f$ (the symbol is called \"nabla\").\n\n\n### Higher-order derivatives\n\nWe saw that velocity is the derivative of position because it describes how position changes over time. Acceleration similarly describes how _velocity_ changes over time, so we'd say that acceleration is the derivative of velocity. We can also say that acceleration is the _second-order_ derivative of position (that is, it is the derivative of its derivative).\n\nThis is the general idea behind higher-order derivatives.\n\n## Gradient descent\n\nOnce you understand derivatives, gradient descent is really, really simple. The basic idea is that we use the derivative of the loss $L(\\theta)$ with respect to $\\theta$ and figure out which way the loss is decreasing, then \"move\" the parameter guess in that direction.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Check Environment\nThis notebook checks that you have correctly created the environment and that all packages needed are installed.", "_____no_output_____" ], [ "## Environment\n\nThe next command should return a line like (Mac/Linux):\n\n /<YOUR-HOME-FOLDER>/anaconda/envs/ztdl/bin/python\n\nor like (Windows 10):\n\n C:\\\\<YOUR-HOME-FOLDER>\\\\Anaconda3\\\\envs\\\\ztdl\\\\python.exe\n\nIn particular you should make sure that you are using the python executable from within the course environment.\n\nIf that's not the case do this:\n\n1. close this notebook\n2. go to the terminal and stop jupyer notebook\n3. make sure that you have activated the environment, you should see a prompt like:\n\n (ztdl) $\n4. (optional) if you don't see that prompt activate the environment:\n - mac/linux:\n \n conda activate ztdl\n\n - windows:\n\n activate ztdl\n5. restart jupyter notebook", "_____no_output_____" ] ], [ [ "import os\nimport sys\nsys.executable", "_____no_output_____" ] ], [ [ "## Python 3.6\n\nThe next line should say that you're using Python 3.6.x from Continuum Analytics. At the time of publication it looks like this (Mac/Linux):\n\n Python 3.6.7 |Anaconda, Inc.| (default, Oct 23 2018, 14:01:38)\n [GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)] on darwin\n Type \"help\", \"copyright\", \"credits\" or \"license\" for more information.\n\nor like this (Windows 10):\n\n Python 3.6.7 |Anaconda, Inc.| (default, Oct 28 2018, 19:44:12) [MSC v.1915 64 bit (AMD64)] on win32\n Type \"help\", \"copyright\", \"credits\" or \"license\" for more information.\n\nbut date and exact version of GCC may change in the future.\n\nIf you see a different version of python, go back to the previous step and make sure you created and activated the environment correctly.", "_____no_output_____" ] ], [ [ "import sys\nsys.version", "_____no_output_____" ] ], [ [ "## Jupyter\n\nCheck that Jupyter is running from within the environment. The next line should look like (Mac/Linux):\n\n /<YOUR-HOME-FOLDER>/anaconda/envs/ztdl/lib/python3.6/site-packages/jupyter.py'\n\nor like this (Windows 10):\n\n C:\\\\Users\\\\paperspace\\\\Anaconda3\\\\envs\\\\ztdl\\\\lib\\\\site-packages\\\\jupyter.py", "_____no_output_____" ] ], [ [ "import jupyter\njupyter.__file__", "_____no_output_____" ] ], [ [ "## Other packages\n\nHere we will check that all the packages are installed and have the correct versions. If everything is ok you should see:\n \n Using TensorFlow backend.\n \n Houston we are go!\n\nIf there's any issue here please make sure you have checked the previous steps and if it's all good please send us a question in the Q&A forum.", "_____no_output_____" ] ], [ [ "import pip\nimport numpy\nimport jupyter\nimport matplotlib\nimport sklearn\nimport scipy\nimport pandas\nimport PIL\nimport seaborn\nimport h5py\nimport tensorflow\nimport keras\n\n\ndef check_version(pkg, version):\n actual = pkg.__version__.split('.')\n if len(actual) == 3:\n actual_major = '.'.join(actual[:2])\n elif len(actual) == 2:\n actual_major = '.'.join(actual)\n else:\n raise NotImplementedError(pkg.__name__ +\n \"actual version :\"+\n pkg.__version__)\n try:\n assert(actual_major == version)\n except Exception as ex:\n print(\"{} {}\\t=> {}\".format(pkg.__name__,\n version,\n pkg.__version__))\n raise ex\n\ncheck_version(pip, '10.0')\ncheck_version(numpy, '1.15')\ncheck_version(matplotlib, '3.0')\ncheck_version(sklearn, '0.20')\ncheck_version(scipy, '1.1')\ncheck_version(pandas, '0.23')\ncheck_version(PIL, '5.3')\ncheck_version(seaborn, '0.9')\ncheck_version(h5py, '2.8')\ncheck_version(tensorflow, '1.11')\ncheck_version(keras, '2.2')\n\nprint(\"Houston we are go!\")", "Houston we are go!\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "ADSL" ]

[ "ADSL" ]

[ "ADSL" ]

[ [ [ "## Preprocessing", "_____no_output_____" ] ], [ [ "# Import our dependencies\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.preprocessing import StandardScaler\nimport pandas as pd\nimport tensorflow as tf\n\n# Import and read the charity_data.csv.\nimport pandas as pd \napplication_df = pd.read_csv(\"charity_data.csv\")\napplication_df.head()", "_____no_output_____" ], [ "# Drop the non-beneficial ID columns, 'EIN' and 'NAME'.\napplication_df = application_df.drop(['EIN', 'NAME'], axis=1)", "_____no_output_____" ], [ "# Determine the number of unique values in each column.\napplication_df.nunique()", "_____no_output_____" ], [ "# Look at APPLICATION_TYPE value counts for binning\nval_count = application_df['APPLICATION_TYPE'].value_counts()\nval_count", "_____no_output_____" ], [ "# Choose a cutoff value and create a list of application types to be replaced\n# use the variable name `application_types_to_replace`\napplication_types_to_replace = list(val_count[val_count<200].index)\n\n# Replace in dataframe\nfor app in application_types_to_replace:\n application_df['APPLICATION_TYPE'] = application_df['APPLICATION_TYPE'].replace(app,\"Other\")\n\n# Check to make sure binning was successful\napplication_df['APPLICATION_TYPE'].value_counts()", "_____no_output_____" ], [ "# Look at CLASSIFICATION value counts for binning\nclass_val_count = application_df['CLASSIFICATION'].value_counts()\nclass_val_count", "_____no_output_____" ], [ "# You may find it helpful to look at CLASSIFICATION value counts >1\nclass_val_count2 = class_val_count[class_val_count>1]\nclass_val_count2", "_____no_output_____" ], [ "# Choose a cutoff value and create a list of classifications to be replaced\n# use the variable name `classifications_to_replace`\nclassifications_to_replace =class_val_count.loc[class_val_count<1000].index\n\n# Replace in dataframe\nfor cls in classifications_to_replace:\n application_df['CLASSIFICATION'] = application_df['CLASSIFICATION'].replace(cls,\"Other\")\n \n# Check to make sure binning was successful\napplication_df['CLASSIFICATION'].value_counts()", "_____no_output_____" ], [ "# Convert categorical data to numeric with `pd.get_dummies`\ndummies_df = pd.get_dummies(application_df)\ndummies_df", "_____no_output_____" ], [ "# Split our preprocessed data into our features and target arrays\nX = dummies_df.drop(['IS_SUCCESSFUL'],1).values\ny = dummies_df['IS_SUCCESSFUL'].values\n\n# Split the preprocessed data into a training and testing dataset\nX_train, X_test, y_train, y_test = train_test_split(X,y,random_state= 42)", "_____no_output_____" ], [ "print(X.shape)\nprint(y.shape)", "(34299, 43)\n(34299,)\n" ], [ "# Create a StandardScaler instances\nscaler = StandardScaler()\n\n# Fit the StandardScaler\nX_scaler = scaler.fit(X_train)\n\n# Scale the data\nX_train_scaled = X_scaler.transform(X_train)\nX_test_scaled = X_scaler.transform(X_test)", "_____no_output_____" ] ], [ [ "## Compile, Train and Evaluate the Model", "_____no_output_____" ] ], [ [ "# Define the model - deep neural net, i.e., the number of input features and hidden nodes for each layer.\ninput_features = len(X_train_scaled[0])\nhidden_nodes_layer1 = 80\nhidden_nodes_layer2 = 30\n#hidden_nodes_layer3 = 60\n\nnn = tf.keras.models.Sequential()\n\n# First hidden layer\nnn.add(tf.keras.layers.Dense(units=hidden_nodes_layer1, input_dim=input_features, activation='relu'))\n\n# Second hidden layer\nnn.add(tf.keras.layers.Dense(units=hidden_nodes_layer2, activation='relu'))\n\n# Third hidden layer\n#nn.add(tf.keras.layers.Dense(units=hidden_nodes_layer3, activation='relu'))\n\n\n# Output layer\nnn.add(tf.keras.layers.Dense(units=1, activation='sigmoid'))\n\n# Check the structure of the model\nnn.summary()", "Model: \"sequential\"\n_________________________________________________________________\n Layer (type) Output Shape Param # \n=================================================================\n dense (Dense) (None, 80) 3520 \n \n dense_1 (Dense) (None, 30) 2430 \n \n dense_2 (Dense) (None, 1) 31 \n \n=================================================================\nTotal params: 5,981\nTrainable params: 5,981\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "# Compile the model\nnn.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])", "_____no_output_____" ], [ "# Train the model\nfit_model = nn.fit(X_train_scaled, y_train, epochs=100)", "Epoch 1/100\n804/804 [==============================] - 6s 4ms/step - loss: 0.5712 - accuracy: 0.7208\nEpoch 2/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5569 - accuracy: 0.7266\nEpoch 3/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5536 - accuracy: 0.7293\nEpoch 4/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5510 - accuracy: 0.7303\nEpoch 5/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5499 - accuracy: 0.7304\nEpoch 6/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5480 - accuracy: 0.7320\nEpoch 7/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5478 - accuracy: 0.7309\nEpoch 8/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5474 - accuracy: 0.7331\nEpoch 9/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5469 - accuracy: 0.7323\nEpoch 10/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5458 - accuracy: 0.7325\nEpoch 11/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5453 - accuracy: 0.7330\nEpoch 12/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5449 - accuracy: 0.7341\nEpoch 13/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5452 - accuracy: 0.7334\nEpoch 14/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5439 - accuracy: 0.7335\nEpoch 15/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5440 - accuracy: 0.7343\nEpoch 16/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5434 - accuracy: 0.7356\nEpoch 17/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5433 - accuracy: 0.7352\nEpoch 18/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5432 - accuracy: 0.7352\nEpoch 19/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5426 - accuracy: 0.7358\nEpoch 20/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5422 - accuracy: 0.7352\nEpoch 21/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5421 - accuracy: 0.7355\nEpoch 22/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5419 - accuracy: 0.7353\nEpoch 23/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5420 - accuracy: 0.7350\nEpoch 24/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5418 - accuracy: 0.7355\nEpoch 25/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5412 - accuracy: 0.7357\nEpoch 26/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5410 - accuracy: 0.7351\nEpoch 27/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5410 - accuracy: 0.7360\nEpoch 28/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5410 - accuracy: 0.7352\nEpoch 29/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5408 - accuracy: 0.7356\nEpoch 30/100\n804/804 [==============================] - 4s 5ms/step - loss: 0.5399 - accuracy: 0.7365\nEpoch 31/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5402 - accuracy: 0.7359\nEpoch 32/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5400 - accuracy: 0.7365\nEpoch 33/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5399 - accuracy: 0.7357\nEpoch 34/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5396 - accuracy: 0.7373\nEpoch 35/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5398 - accuracy: 0.7365\nEpoch 36/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5396 - accuracy: 0.7376\nEpoch 37/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5394 - accuracy: 0.7365\nEpoch 38/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5391 - accuracy: 0.7371\nEpoch 39/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5391 - accuracy: 0.7362\nEpoch 40/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5392 - accuracy: 0.7369\nEpoch 41/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5389 - accuracy: 0.7372\nEpoch 42/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5386 - accuracy: 0.7381\nEpoch 43/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5381 - accuracy: 0.7374\nEpoch 44/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5384 - accuracy: 0.7375\nEpoch 45/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5381 - accuracy: 0.7382\nEpoch 46/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5383 - accuracy: 0.7387\nEpoch 47/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5381 - accuracy: 0.7379\nEpoch 48/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5378 - accuracy: 0.7370\nEpoch 49/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5374 - accuracy: 0.7376\nEpoch 50/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5376 - accuracy: 0.7380\nEpoch 51/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5373 - accuracy: 0.7378\nEpoch 52/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5377 - accuracy: 0.7378\nEpoch 53/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5371 - accuracy: 0.7385\nEpoch 54/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5370 - accuracy: 0.7377\nEpoch 55/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5371 - accuracy: 0.7388\nEpoch 56/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5369 - accuracy: 0.7391\nEpoch 57/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5371 - accuracy: 0.7386\nEpoch 58/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5368 - accuracy: 0.7392\nEpoch 59/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5364 - accuracy: 0.7386\nEpoch 60/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5367 - accuracy: 0.7389\nEpoch 61/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5362 - accuracy: 0.7396\nEpoch 62/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5364 - accuracy: 0.7386\nEpoch 63/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5363 - accuracy: 0.7391\nEpoch 64/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5363 - accuracy: 0.7387\nEpoch 65/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5362 - accuracy: 0.7396\nEpoch 66/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7388\nEpoch 67/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5360 - accuracy: 0.7391\nEpoch 68/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5356 - accuracy: 0.7392\nEpoch 69/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5360 - accuracy: 0.7385\nEpoch 70/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5355 - accuracy: 0.7391\nEpoch 71/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5355 - accuracy: 0.7397\nEpoch 72/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5358 - accuracy: 0.7391\nEpoch 73/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5357 - accuracy: 0.7401\nEpoch 74/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5356 - accuracy: 0.7388\nEpoch 75/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5356 - accuracy: 0.7394\nEpoch 76/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5356 - accuracy: 0.7393\nEpoch 77/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5352 - accuracy: 0.7394\nEpoch 78/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5352 - accuracy: 0.7400\nEpoch 79/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5353 - accuracy: 0.7398\nEpoch 80/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5352 - accuracy: 0.7393\nEpoch 81/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5351 - accuracy: 0.7396\nEpoch 82/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5349 - accuracy: 0.7391\nEpoch 83/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5348 - accuracy: 0.7395\nEpoch 84/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5346 - accuracy: 0.7397\nEpoch 85/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5349 - accuracy: 0.7394\nEpoch 86/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5347 - accuracy: 0.7397\nEpoch 87/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5347 - accuracy: 0.7397\nEpoch 88/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5348 - accuracy: 0.7400\nEpoch 89/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5348 - accuracy: 0.7408\nEpoch 90/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5345 - accuracy: 0.7397\nEpoch 91/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5345 - accuracy: 0.7397\nEpoch 92/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5341 - accuracy: 0.7402\nEpoch 93/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7406\nEpoch 94/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7398\nEpoch 95/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5343 - accuracy: 0.7411\nEpoch 96/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5346 - accuracy: 0.7396\nEpoch 97/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5339 - accuracy: 0.7392\nEpoch 98/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7401\nEpoch 99/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7396\nEpoch 100/100\n804/804 [==============================] - 3s 4ms/step - loss: 0.5339 - accuracy: 0.7398\n" ], [ "# Evaluate the model using the test data\nmodel_loss, model_accuracy = nn.evaluate(X_test_scaled,y_test,verbose=2)\nprint(f\"Loss: {model_loss}, Accuracy: {model_accuracy}\")", "268/268 - 1s - loss: 0.5639 - accuracy: 0.7277 - 735ms/epoch - 3ms/step\nLoss: 0.5639024972915649, Accuracy: 0.7276967763900757\n" ], [ "# Export our model to HDF5 file\nnn.save(\"AlphabetSoupCharity.h5\") ", "_____no_output_____" ] ], [ [ "**Optimization 1**\n", "_____no_output_____" ] ], [ [ "# Define the model - deep neural net, i.e., the number of input features and hidden nodes for each layer.\ninput_features = len(X_train_scaled[0])\nhidden_nodes_layer1 = 43\nhidden_nodes_layer2 = 25\n#hidden_nodes_layer3 = 15\n\nnn2 = tf.keras.models.Sequential()\n\n# First hidden layer\nnn2.add(tf.keras.layers.Dense(units=hidden_nodes_layer1, input_dim=input_features, activation='relu'))\n\n# Second hidden layer\nnn2.add(tf.keras.layers.Dense(units=hidden_nodes_layer2, activation='relu'))\n\n# Third hidden layer\n#nn.add(tf.keras.layers.Dense(units=hidden_nodes_layer3, activation='relu'))\n\n\n# Output layer\nnn2.add(tf.keras.layers.Dense(units=1, activation='sigmoid'))\n\n# Check the structure of the model\nnn2.summary()", "Model: \"sequential_1\"\n_________________________________________________________________\n Layer (type) Output Shape Param # \n=================================================================\n dense_3 (Dense) (None, 43) 1892 \n \n dense_4 (Dense) (None, 25) 1100 \n \n dense_5 (Dense) (None, 1) 26 \n \n=================================================================\nTotal params: 3,018\nTrainable params: 3,018\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "# Compile the model\nnn2.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])", "_____no_output_____" ], [ "# Train the model\nfit_model = nn2.fit(X_train_scaled, y_train, epochs=200)", "Epoch 1/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5765 - accuracy: 0.7133\nEpoch 2/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5556 - accuracy: 0.7291\nEpoch 3/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5529 - accuracy: 0.7300\nEpoch 4/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5502 - accuracy: 0.7303\nEpoch 5/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5501 - accuracy: 0.7304\nEpoch 6/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5483 - accuracy: 0.7320\nEpoch 7/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5479 - accuracy: 0.7312\nEpoch 8/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5479 - accuracy: 0.7320\nEpoch 9/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5471 - accuracy: 0.7326\nEpoch 10/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5465 - accuracy: 0.7324\nEpoch 11/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5452 - accuracy: 0.7329\nEpoch 12/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5456 - accuracy: 0.7337\nEpoch 13/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5450 - accuracy: 0.7342\nEpoch 14/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5449 - accuracy: 0.7331\nEpoch 15/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5444 - accuracy: 0.7332\nEpoch 16/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5445 - accuracy: 0.7333\nEpoch 17/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5435 - accuracy: 0.7346\nEpoch 18/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5434 - accuracy: 0.7346\nEpoch 19/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5432 - accuracy: 0.7338\nEpoch 20/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5431 - accuracy: 0.7353\nEpoch 21/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5428 - accuracy: 0.7347\nEpoch 22/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5428 - accuracy: 0.7352\nEpoch 23/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5422 - accuracy: 0.7347\nEpoch 24/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5426 - accuracy: 0.7349\nEpoch 25/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5424 - accuracy: 0.7362\nEpoch 26/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5423 - accuracy: 0.7345\nEpoch 27/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5417 - accuracy: 0.7363\nEpoch 28/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5419 - accuracy: 0.7358\nEpoch 29/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5418 - accuracy: 0.7355\nEpoch 30/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5414 - accuracy: 0.7350\nEpoch 31/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5411 - accuracy: 0.7355\nEpoch 32/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5407 - accuracy: 0.7353\nEpoch 33/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5406 - accuracy: 0.7371\nEpoch 34/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5409 - accuracy: 0.7364\nEpoch 35/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5406 - accuracy: 0.7362\nEpoch 36/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5403 - accuracy: 0.7364\nEpoch 37/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5402 - accuracy: 0.7360\nEpoch 38/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5405 - accuracy: 0.7365\nEpoch 39/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5401 - accuracy: 0.7369\nEpoch 40/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5401 - accuracy: 0.7364\nEpoch 41/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5403 - accuracy: 0.7366\nEpoch 42/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5396 - accuracy: 0.7362\nEpoch 43/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5397 - accuracy: 0.7358\nEpoch 44/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5392 - accuracy: 0.7365\nEpoch 45/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5395 - accuracy: 0.7370\nEpoch 46/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5396 - accuracy: 0.7374\nEpoch 47/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5392 - accuracy: 0.7359\nEpoch 48/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5391 - accuracy: 0.7369\nEpoch 49/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5391 - accuracy: 0.7369\nEpoch 50/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5389 - accuracy: 0.7381\nEpoch 51/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5390 - accuracy: 0.7371\nEpoch 52/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5392 - accuracy: 0.7378\nEpoch 53/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5390 - accuracy: 0.7373\nEpoch 54/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5388 - accuracy: 0.7374\nEpoch 55/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5387 - accuracy: 0.7372\nEpoch 56/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5388 - accuracy: 0.7363\nEpoch 57/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5389 - accuracy: 0.7354\nEpoch 58/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5386 - accuracy: 0.7375\nEpoch 59/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5384 - accuracy: 0.7376\nEpoch 60/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5379 - accuracy: 0.7375\nEpoch 61/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5381 - accuracy: 0.7382\nEpoch 62/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5380 - accuracy: 0.7381\nEpoch 63/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5379 - accuracy: 0.7371\nEpoch 64/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5381 - accuracy: 0.7380\nEpoch 65/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5380 - accuracy: 0.7383\nEpoch 66/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5381 - accuracy: 0.7378\nEpoch 67/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5378 - accuracy: 0.7379\nEpoch 68/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5380 - accuracy: 0.7382\nEpoch 69/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5378 - accuracy: 0.7382\nEpoch 70/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5379 - accuracy: 0.7381\nEpoch 71/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5375 - accuracy: 0.7378\nEpoch 72/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5374 - accuracy: 0.7385\nEpoch 73/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5375 - accuracy: 0.7372\nEpoch 74/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5374 - accuracy: 0.7385\nEpoch 75/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5373 - accuracy: 0.7374\nEpoch 76/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5370 - accuracy: 0.7387\nEpoch 77/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5372 - accuracy: 0.7391\nEpoch 78/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5372 - accuracy: 0.7388\nEpoch 79/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5372 - accuracy: 0.7390\nEpoch 80/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5369 - accuracy: 0.7386\nEpoch 81/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5371 - accuracy: 0.7389\nEpoch 82/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5369 - accuracy: 0.7382\nEpoch 83/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5368 - accuracy: 0.7391\nEpoch 84/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5371 - accuracy: 0.7379\nEpoch 85/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5366 - accuracy: 0.7374\nEpoch 86/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5370 - accuracy: 0.7371\nEpoch 87/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5364 - accuracy: 0.7390\nEpoch 88/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5368 - accuracy: 0.7386\nEpoch 89/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5366 - accuracy: 0.7390\nEpoch 90/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5363 - accuracy: 0.7394\nEpoch 91/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7385\nEpoch 92/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5366 - accuracy: 0.7393\nEpoch 93/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5366 - accuracy: 0.7389\nEpoch 94/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5362 - accuracy: 0.7390\nEpoch 95/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5362 - accuracy: 0.7391\nEpoch 96/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5362 - accuracy: 0.7386\nEpoch 97/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5362 - accuracy: 0.7386\nEpoch 98/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5359 - accuracy: 0.7388\nEpoch 99/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5365 - accuracy: 0.7385\nEpoch 100/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5364 - accuracy: 0.7388\nEpoch 101/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5363 - accuracy: 0.7390\nEpoch 102/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5359 - accuracy: 0.7375\nEpoch 103/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7385\nEpoch 104/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5360 - accuracy: 0.7389\nEpoch 105/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5356 - accuracy: 0.7393\nEpoch 106/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5358 - accuracy: 0.7392\nEpoch 107/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5357 - accuracy: 0.7386\nEpoch 108/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7393\nEpoch 109/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5359 - accuracy: 0.7388\nEpoch 110/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5354 - accuracy: 0.7387\nEpoch 111/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5358 - accuracy: 0.7396\nEpoch 112/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5355 - accuracy: 0.7398\nEpoch 113/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5360 - accuracy: 0.7397\nEpoch 114/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5356 - accuracy: 0.7381\nEpoch 115/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5357 - accuracy: 0.7390\nEpoch 116/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5352 - accuracy: 0.7388\nEpoch 117/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5353 - accuracy: 0.7391\nEpoch 118/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5356 - accuracy: 0.7393\nEpoch 119/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5358 - accuracy: 0.7394\nEpoch 120/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5350 - accuracy: 0.7397\nEpoch 121/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5354 - accuracy: 0.7384\nEpoch 122/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5353 - accuracy: 0.7401\nEpoch 123/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5352 - accuracy: 0.7393\nEpoch 124/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5352 - accuracy: 0.7404\nEpoch 125/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5349 - accuracy: 0.7392\nEpoch 126/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5349 - accuracy: 0.7404\nEpoch 127/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5346 - accuracy: 0.7402\nEpoch 128/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5352 - accuracy: 0.7403\nEpoch 129/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5350 - accuracy: 0.7398\nEpoch 130/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5350 - accuracy: 0.7398\nEpoch 131/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5344 - accuracy: 0.7394\nEpoch 132/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5349 - accuracy: 0.7402\nEpoch 133/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5348 - accuracy: 0.7397\nEpoch 134/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5352 - accuracy: 0.7392\nEpoch 135/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5349 - accuracy: 0.7408\nEpoch 136/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5348 - accuracy: 0.7403\nEpoch 137/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5349 - accuracy: 0.7401\nEpoch 138/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5350 - accuracy: 0.7404\nEpoch 139/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5345 - accuracy: 0.7397\nEpoch 140/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5350 - accuracy: 0.7396\nEpoch 141/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5342 - accuracy: 0.7399\nEpoch 142/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5347 - accuracy: 0.7405\nEpoch 143/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5347 - accuracy: 0.7389\nEpoch 144/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5347 - accuracy: 0.7405\nEpoch 145/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5348 - accuracy: 0.7396\nEpoch 146/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5344 - accuracy: 0.7407\nEpoch 147/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5347 - accuracy: 0.7402\nEpoch 148/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5346 - accuracy: 0.7401\nEpoch 149/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7399\nEpoch 150/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7398\nEpoch 151/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5346 - accuracy: 0.7400\nEpoch 152/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5346 - accuracy: 0.7414\nEpoch 153/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7408\nEpoch 154/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5345 - accuracy: 0.7395\nEpoch 155/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7400\nEpoch 156/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5341 - accuracy: 0.7400\nEpoch 157/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5346 - accuracy: 0.7404\nEpoch 158/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5342 - accuracy: 0.7407\nEpoch 159/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7406\nEpoch 160/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7408\nEpoch 161/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5341 - accuracy: 0.7414\nEpoch 162/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5341 - accuracy: 0.7405\nEpoch 163/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5341 - accuracy: 0.7398\nEpoch 164/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5345 - accuracy: 0.7401\nEpoch 165/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5338 - accuracy: 0.7411\nEpoch 166/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5339 - accuracy: 0.7408\nEpoch 167/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7404\nEpoch 168/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5341 - accuracy: 0.7392\nEpoch 169/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7400\nEpoch 170/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7400\nEpoch 171/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7409\nEpoch 172/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5342 - accuracy: 0.7396\nEpoch 173/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5342 - accuracy: 0.7405\nEpoch 174/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5339 - accuracy: 0.7404\nEpoch 175/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5339 - accuracy: 0.7404\nEpoch 176/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7408\nEpoch 177/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5343 - accuracy: 0.7395\nEpoch 178/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7408\nEpoch 179/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7402\nEpoch 180/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7406\nEpoch 181/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7410\nEpoch 182/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7401\nEpoch 183/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5339 - accuracy: 0.7399\nEpoch 184/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5336 - accuracy: 0.7407\nEpoch 185/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7408\nEpoch 186/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5338 - accuracy: 0.7412\nEpoch 187/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5334 - accuracy: 0.7405\nEpoch 188/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5336 - accuracy: 0.7407\nEpoch 189/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7416\nEpoch 190/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7409\nEpoch 191/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7408\nEpoch 192/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5336 - accuracy: 0.7405\nEpoch 193/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5332 - accuracy: 0.7399\nEpoch 194/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5334 - accuracy: 0.7404\nEpoch 195/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5335 - accuracy: 0.7408\nEpoch 196/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7407\nEpoch 197/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5337 - accuracy: 0.7414\nEpoch 198/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7406\nEpoch 199/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5333 - accuracy: 0.7411\nEpoch 200/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5334 - accuracy: 0.7413\n" ], [ "# Evaluate the model using the test data\nmodel_loss, model_accuracy = nn2.evaluate(X_test_scaled,y_test,verbose=2)\nprint(f\"Loss: {model_loss}, Accuracy: {model_accuracy}\")", "268/268 - 1s - loss: 0.5651 - accuracy: 0.7263 - 711ms/epoch - 3ms/step\nLoss: 0.5651004910469055, Accuracy: 0.7262973785400391\n" ], [ "# Export our model to HDF5 file\nnn2.save(\"AlphabetSoupCharity_optimize_1.h5\") ", "_____no_output_____" ] ], [ [ "**Optimization 2**", "_____no_output_____" ] ], [ [ "# Define the model - deep neural net, i.e., the number of input features and hidden nodes for each layer.\ninput_features = len(X_train_scaled[0])\nhidden_nodes_layer1 = 20\nhidden_nodes_layer2 = 40\nhidden_nodes_layer3 = 80\n\nnn3 = tf.keras.models.Sequential()\n\n# First hidden layer\nnn3.add(tf.keras.layers.Dense(units=hidden_nodes_layer1, input_dim=input_features, activation='relu'))\n\n# Second hidden layer\nnn3.add(tf.keras.layers.Dense(units=hidden_nodes_layer2, activation='relu'))\n\n# Third hidden layer\nnn3.add(tf.keras.layers.Dense(units=hidden_nodes_layer3, activation='relu'))\n\n\n# Output layer\nnn3.add(tf.keras.layers.Dense(units=1, activation='sigmoid'))\n\n# Check the structure of the model\nnn3.summary()", "Model: \"sequential_2\"\n_________________________________________________________________\n Layer (type) Output Shape Param # \n=================================================================\n dense_6 (Dense) (None, 20) 880 \n \n dense_7 (Dense) (None, 40) 840 \n \n dense_8 (Dense) (None, 80) 3280 \n \n dense_9 (Dense) (None, 1) 81 \n \n=================================================================\nTotal params: 5,081\nTrainable params: 5,081\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "# Compile the model\nnn3.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])", "_____no_output_____" ], [ "# Train the model\nfit_model = nn3.fit(X_train_scaled, y_train, epochs=200)", "Epoch 1/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5725 - accuracy: 0.7208\nEpoch 2/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5562 - accuracy: 0.7286\nEpoch 3/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5533 - accuracy: 0.7282\nEpoch 4/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5513 - accuracy: 0.7301\nEpoch 5/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5496 - accuracy: 0.7310\nEpoch 6/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5487 - accuracy: 0.7313\nEpoch 7/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5481 - accuracy: 0.7322\nEpoch 8/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5475 - accuracy: 0.7318\nEpoch 9/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5466 - accuracy: 0.7338\nEpoch 10/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5458 - accuracy: 0.7333\nEpoch 11/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5454 - accuracy: 0.7338\nEpoch 12/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5449 - accuracy: 0.7329\nEpoch 13/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5450 - accuracy: 0.7344\nEpoch 14/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5440 - accuracy: 0.7359\nEpoch 15/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5445 - accuracy: 0.7355\nEpoch 16/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5440 - accuracy: 0.7336\nEpoch 17/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5435 - accuracy: 0.7348\nEpoch 18/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5432 - accuracy: 0.7350\nEpoch 19/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5430 - accuracy: 0.7345\nEpoch 20/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5425 - accuracy: 0.7348\nEpoch 21/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5424 - accuracy: 0.7357\nEpoch 22/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5422 - accuracy: 0.7357\nEpoch 23/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5421 - accuracy: 0.7358\nEpoch 24/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5422 - accuracy: 0.7355\nEpoch 25/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5416 - accuracy: 0.7355\nEpoch 26/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5413 - accuracy: 0.7360\nEpoch 27/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5411 - accuracy: 0.7364\nEpoch 28/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5413 - accuracy: 0.7354\nEpoch 29/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5411 - accuracy: 0.7364\nEpoch 30/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5409 - accuracy: 0.7360\nEpoch 31/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5408 - accuracy: 0.7365\nEpoch 32/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5402 - accuracy: 0.7362\nEpoch 33/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5400 - accuracy: 0.7365\nEpoch 34/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5404 - accuracy: 0.7358\nEpoch 35/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5397 - accuracy: 0.7385\nEpoch 36/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5399 - accuracy: 0.7366\nEpoch 37/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5398 - accuracy: 0.7373\nEpoch 38/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5392 - accuracy: 0.7375\nEpoch 39/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5390 - accuracy: 0.7381\nEpoch 40/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5388 - accuracy: 0.7381\nEpoch 41/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5388 - accuracy: 0.7387\nEpoch 42/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5385 - accuracy: 0.7371\nEpoch 43/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5386 - accuracy: 0.7370\nEpoch 44/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5387 - accuracy: 0.7381\nEpoch 45/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5385 - accuracy: 0.7369\nEpoch 46/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5380 - accuracy: 0.7388\nEpoch 47/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5386 - accuracy: 0.7378\nEpoch 48/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5382 - accuracy: 0.7378\nEpoch 49/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5383 - accuracy: 0.7373\nEpoch 50/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5378 - accuracy: 0.7378\nEpoch 51/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5374 - accuracy: 0.7384\nEpoch 52/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5374 - accuracy: 0.7378\nEpoch 53/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5383 - accuracy: 0.7377\nEpoch 54/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5376 - accuracy: 0.7383\nEpoch 55/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5375 - accuracy: 0.7379\nEpoch 56/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5371 - accuracy: 0.7383\nEpoch 57/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5368 - accuracy: 0.7398\nEpoch 58/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5369 - accuracy: 0.7389\nEpoch 59/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5372 - accuracy: 0.7373\nEpoch 60/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5368 - accuracy: 0.7392\nEpoch 61/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5364 - accuracy: 0.7394\nEpoch 62/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5367 - accuracy: 0.7384\nEpoch 63/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5364 - accuracy: 0.7392\nEpoch 64/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5364 - accuracy: 0.7392\nEpoch 65/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5363 - accuracy: 0.7396\nEpoch 66/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5365 - accuracy: 0.7389\nEpoch 67/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7392\nEpoch 68/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7396\nEpoch 69/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5361 - accuracy: 0.7387\nEpoch 70/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5359 - accuracy: 0.7388\nEpoch 71/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5355 - accuracy: 0.7399\nEpoch 72/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5360 - accuracy: 0.7387\nEpoch 73/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5361 - accuracy: 0.7396\nEpoch 74/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5359 - accuracy: 0.7393\nEpoch 75/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5354 - accuracy: 0.7394\nEpoch 76/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5354 - accuracy: 0.7399\nEpoch 77/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5354 - accuracy: 0.7388\nEpoch 78/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5360 - accuracy: 0.7395\nEpoch 79/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5353 - accuracy: 0.7400\nEpoch 80/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5350 - accuracy: 0.7395\nEpoch 81/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5353 - accuracy: 0.7394\nEpoch 82/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5351 - accuracy: 0.7392\nEpoch 83/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5347 - accuracy: 0.7406\nEpoch 84/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5348 - accuracy: 0.7406\nEpoch 85/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5350 - accuracy: 0.7392\nEpoch 86/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5344 - accuracy: 0.7406\nEpoch 87/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5348 - accuracy: 0.7406\nEpoch 88/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5348 - accuracy: 0.7396\nEpoch 89/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5349 - accuracy: 0.7395\nEpoch 90/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7397\nEpoch 91/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5344 - accuracy: 0.7401\nEpoch 92/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5344 - accuracy: 0.7406\nEpoch 93/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5347 - accuracy: 0.7402\nEpoch 94/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5345 - accuracy: 0.7390\nEpoch 95/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5346 - accuracy: 0.7399\nEpoch 96/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5345 - accuracy: 0.7397\nEpoch 97/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5343 - accuracy: 0.7404\nEpoch 98/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5340 - accuracy: 0.7404\nEpoch 99/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5338 - accuracy: 0.7403\nEpoch 100/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5345 - accuracy: 0.7407\nEpoch 101/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5343 - accuracy: 0.7411\nEpoch 102/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7400\nEpoch 103/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5339 - accuracy: 0.7404\nEpoch 104/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5342 - accuracy: 0.7402\nEpoch 105/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5338 - accuracy: 0.7405\nEpoch 106/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5340 - accuracy: 0.7404\nEpoch 107/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5345 - accuracy: 0.7402\nEpoch 108/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5337 - accuracy: 0.7400\nEpoch 109/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5335 - accuracy: 0.7406\nEpoch 110/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5335 - accuracy: 0.7407\nEpoch 111/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7400\nEpoch 112/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7396\nEpoch 113/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5332 - accuracy: 0.7409\nEpoch 114/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5332 - accuracy: 0.7404\nEpoch 115/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5334 - accuracy: 0.7409\nEpoch 116/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5334 - accuracy: 0.7404\nEpoch 117/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5332 - accuracy: 0.7407\nEpoch 118/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5336 - accuracy: 0.7400\nEpoch 119/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5333 - accuracy: 0.7411\nEpoch 120/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5338 - accuracy: 0.7402\nEpoch 121/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5340 - accuracy: 0.7397\nEpoch 122/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5329 - accuracy: 0.7405\nEpoch 123/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5332 - accuracy: 0.7413\nEpoch 124/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5331 - accuracy: 0.7410\nEpoch 125/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5332 - accuracy: 0.7407\nEpoch 126/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5333 - accuracy: 0.7410\nEpoch 127/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5333 - accuracy: 0.7409\nEpoch 128/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5333 - accuracy: 0.7403\nEpoch 129/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5330 - accuracy: 0.7410\nEpoch 130/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5338 - accuracy: 0.7405\nEpoch 131/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5331 - accuracy: 0.7407\nEpoch 132/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5330 - accuracy: 0.7410\nEpoch 133/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5331 - accuracy: 0.7400\nEpoch 134/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5328 - accuracy: 0.7413\nEpoch 135/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5330 - accuracy: 0.7405\nEpoch 136/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5332 - accuracy: 0.7407\nEpoch 137/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5330 - accuracy: 0.7405\nEpoch 138/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5329 - accuracy: 0.7406\nEpoch 139/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5327 - accuracy: 0.7406\nEpoch 140/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5328 - accuracy: 0.7407\nEpoch 141/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5329 - accuracy: 0.7405\nEpoch 142/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5333 - accuracy: 0.7402\nEpoch 143/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5332 - accuracy: 0.7413\nEpoch 144/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5324 - accuracy: 0.7412\nEpoch 145/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5327 - accuracy: 0.7404\nEpoch 146/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5325 - accuracy: 0.7411\nEpoch 147/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5326 - accuracy: 0.7408\nEpoch 148/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5325 - accuracy: 0.7410\nEpoch 149/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5330 - accuracy: 0.7406\nEpoch 150/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5327 - accuracy: 0.7409\nEpoch 151/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5327 - accuracy: 0.7411\nEpoch 152/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5328 - accuracy: 0.7402\nEpoch 153/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5323 - accuracy: 0.7409\nEpoch 154/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5318 - accuracy: 0.7419\nEpoch 155/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5323 - accuracy: 0.7416\nEpoch 156/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5325 - accuracy: 0.7413\nEpoch 157/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5323 - accuracy: 0.7414\nEpoch 158/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5327 - accuracy: 0.7406\nEpoch 159/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5321 - accuracy: 0.7409\nEpoch 160/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5324 - accuracy: 0.7406\nEpoch 161/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5324 - accuracy: 0.7409\nEpoch 162/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5323 - accuracy: 0.7412\nEpoch 163/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5321 - accuracy: 0.7404\nEpoch 164/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5323 - accuracy: 0.7411\nEpoch 165/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5323 - accuracy: 0.7415\nEpoch 166/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5322 - accuracy: 0.7405\nEpoch 167/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5318 - accuracy: 0.7413\nEpoch 168/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5323 - accuracy: 0.7415\nEpoch 169/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5318 - accuracy: 0.7405\nEpoch 170/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5319 - accuracy: 0.7409\nEpoch 171/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5320 - accuracy: 0.7414\nEpoch 172/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5327 - accuracy: 0.7408\nEpoch 173/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5317 - accuracy: 0.7411\nEpoch 174/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5320 - accuracy: 0.7412\nEpoch 175/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5315 - accuracy: 0.7411\nEpoch 176/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5318 - accuracy: 0.7409\nEpoch 177/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5323 - accuracy: 0.7404\nEpoch 178/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5321 - accuracy: 0.7405\nEpoch 179/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5321 - accuracy: 0.7403\nEpoch 180/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5320 - accuracy: 0.7415\nEpoch 181/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5319 - accuracy: 0.7414\nEpoch 182/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5318 - accuracy: 0.7411\nEpoch 183/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5318 - accuracy: 0.7409\nEpoch 184/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5322 - accuracy: 0.7413\nEpoch 185/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5314 - accuracy: 0.7421\nEpoch 186/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5319 - accuracy: 0.7406\nEpoch 187/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5316 - accuracy: 0.7407\nEpoch 188/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5329 - accuracy: 0.7408\nEpoch 189/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5319 - accuracy: 0.7419\nEpoch 190/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5316 - accuracy: 0.7412\nEpoch 191/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5314 - accuracy: 0.7413\nEpoch 192/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5317 - accuracy: 0.7422\nEpoch 193/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5321 - accuracy: 0.7412\nEpoch 194/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5318 - accuracy: 0.7406\nEpoch 195/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5313 - accuracy: 0.7405\nEpoch 196/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5315 - accuracy: 0.7416\nEpoch 197/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5313 - accuracy: 0.7410\nEpoch 198/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5313 - accuracy: 0.7413\nEpoch 199/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5317 - accuracy: 0.7414\nEpoch 200/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5322 - accuracy: 0.7413\n" ], [ "# Evaluate the model using the test data\nmodel_loss, model_accuracy = nn3.evaluate(X_test_scaled,y_test,verbose=2)\nprint(f\"Loss: {model_loss}, Accuracy: {model_accuracy}\")", "268/268 - 1s - loss: 0.5761 - accuracy: 0.7322 - 742ms/epoch - 3ms/step\nLoss: 0.576053261756897, Accuracy: 0.7322449088096619\n" ], [ "# Export our model to HDF5 file\nnn3.save(\"AlphabetSoupCharity_optimize_2.h5\") ", "_____no_output_____" ], [ "# Define the model - deep neural net, i.e., the number of input features and hidden nodes for each layer.\ninput_features = len(X_train_scaled[0])\nhidden_nodes_layer1 = 60\nhidden_nodes_layer2 = 60\nhidden_nodes_layer3 = 40\nhidden_nodes_layer4 = 20\n\nnn4 = tf.keras.models.Sequential()\n\n# First hidden layer\nnn4.add(tf.keras.layers.Dense(units=hidden_nodes_layer1, input_dim=input_features, activation='relu'))\n\n# Second hidden layer\nnn4.add(tf.keras.layers.Dense(units=hidden_nodes_layer2, activation='relu'))\n\n# Third hidden layer\nnn4.add(tf.keras.layers.Dense(units=hidden_nodes_layer3, activation='relu'))\n\n# Forth hidden layer\nnn4.add(tf.keras.layers.Dense(units=hidden_nodes_layer4, activation='relu'))\n\n# Output layer\nnn4.add(tf.keras.layers.Dense(units=1, activation='sigmoid'))\n\n# Check the structure of the model\nnn4.summary()", "Model: \"sequential_3\"\n_________________________________________________________________\n Layer (type) Output Shape Param # \n=================================================================\n dense_10 (Dense) (None, 60) 2640 \n \n dense_11 (Dense) (None, 60) 3660 \n \n dense_12 (Dense) (None, 40) 2440 \n \n dense_13 (Dense) (None, 20) 820 \n \n dense_14 (Dense) (None, 1) 21 \n \n=================================================================\nTotal params: 9,581\nTrainable params: 9,581\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "# Compile the model\nnn4.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])", "_____no_output_____" ], [ "# Train the model\nfit_model = nn4.fit(X_train_scaled, y_train, epochs=200)", "Epoch 1/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5713 - accuracy: 0.7197\nEpoch 2/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5558 - accuracy: 0.7264\nEpoch 3/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5527 - accuracy: 0.7309\nEpoch 4/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5503 - accuracy: 0.7327\nEpoch 5/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5492 - accuracy: 0.7311\nEpoch 6/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5477 - accuracy: 0.7318\nEpoch 7/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5467 - accuracy: 0.7329\nEpoch 8/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5464 - accuracy: 0.7341\nEpoch 9/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5451 - accuracy: 0.7333\nEpoch 10/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5449 - accuracy: 0.7344\nEpoch 11/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5445 - accuracy: 0.7340\nEpoch 12/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5445 - accuracy: 0.7341\nEpoch 13/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5434 - accuracy: 0.7356\nEpoch 14/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5433 - accuracy: 0.7345\nEpoch 15/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5424 - accuracy: 0.7358\nEpoch 16/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5425 - accuracy: 0.7345\nEpoch 17/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5421 - accuracy: 0.7348\nEpoch 18/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5413 - accuracy: 0.7360\nEpoch 19/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5418 - accuracy: 0.7361\nEpoch 20/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5409 - accuracy: 0.7369\nEpoch 21/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5404 - accuracy: 0.7359\nEpoch 22/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5405 - accuracy: 0.7367\nEpoch 23/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5406 - accuracy: 0.7365\nEpoch 24/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5404 - accuracy: 0.7367\nEpoch 25/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5400 - accuracy: 0.7362\nEpoch 26/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5394 - accuracy: 0.7367\nEpoch 27/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5395 - accuracy: 0.7362\nEpoch 28/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5388 - accuracy: 0.7372\nEpoch 29/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5393 - accuracy: 0.7373\nEpoch 30/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5388 - accuracy: 0.7372\nEpoch 31/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5386 - accuracy: 0.7379\nEpoch 32/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5382 - accuracy: 0.7374\nEpoch 33/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5383 - accuracy: 0.7385\nEpoch 34/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5380 - accuracy: 0.7381\nEpoch 35/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5373 - accuracy: 0.7378\nEpoch 36/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5374 - accuracy: 0.7385\nEpoch 37/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5382 - accuracy: 0.7376\nEpoch 38/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5371 - accuracy: 0.7384\nEpoch 39/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5368 - accuracy: 0.7386\nEpoch 40/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5366 - accuracy: 0.7382\nEpoch 41/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5367 - accuracy: 0.7381\nEpoch 42/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5365 - accuracy: 0.7390\nEpoch 43/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5361 - accuracy: 0.7388\nEpoch 44/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5364 - accuracy: 0.7386\nEpoch 45/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5366 - accuracy: 0.7387\nEpoch 46/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5362 - accuracy: 0.7388\nEpoch 47/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5357 - accuracy: 0.7383\nEpoch 48/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5359 - accuracy: 0.7384\nEpoch 49/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5360 - accuracy: 0.7391\nEpoch 50/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5361 - accuracy: 0.7383\nEpoch 51/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5354 - accuracy: 0.7395\nEpoch 52/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5350 - accuracy: 0.7369\nEpoch 53/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5351 - accuracy: 0.7400\nEpoch 54/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5349 - accuracy: 0.7390\nEpoch 55/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5347 - accuracy: 0.7383\nEpoch 56/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5349 - accuracy: 0.7394\nEpoch 57/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5346 - accuracy: 0.7392\nEpoch 58/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5350 - accuracy: 0.7393\nEpoch 59/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5346 - accuracy: 0.7399\nEpoch 60/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5344 - accuracy: 0.7396\nEpoch 61/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5354 - accuracy: 0.7391\nEpoch 62/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5346 - accuracy: 0.7385\nEpoch 63/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5340 - accuracy: 0.7401\nEpoch 64/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5343 - accuracy: 0.7396\nEpoch 65/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5336 - accuracy: 0.7402\nEpoch 66/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5339 - accuracy: 0.7397\nEpoch 67/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5338 - accuracy: 0.7391\nEpoch 68/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5342 - accuracy: 0.7405\nEpoch 69/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5340 - accuracy: 0.7395\nEpoch 70/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5339 - accuracy: 0.7398\nEpoch 71/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5338 - accuracy: 0.7402\nEpoch 72/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5338 - accuracy: 0.7403\nEpoch 73/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5340 - accuracy: 0.7405\nEpoch 74/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5331 - accuracy: 0.7405\nEpoch 75/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5336 - accuracy: 0.7409\nEpoch 76/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5335 - accuracy: 0.7397\nEpoch 77/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5331 - accuracy: 0.7400\nEpoch 78/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5334 - accuracy: 0.7397\nEpoch 79/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5330 - accuracy: 0.7403\nEpoch 80/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5330 - accuracy: 0.7398\nEpoch 81/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5327 - accuracy: 0.7406\nEpoch 82/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5328 - accuracy: 0.7401\nEpoch 83/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5331 - accuracy: 0.7401\nEpoch 84/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5321 - accuracy: 0.7399\nEpoch 85/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5328 - accuracy: 0.7410\nEpoch 86/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5329 - accuracy: 0.7403\nEpoch 87/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5342 - accuracy: 0.7405\nEpoch 88/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5320 - accuracy: 0.7407\nEpoch 89/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5342 - accuracy: 0.7401\nEpoch 90/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5322 - accuracy: 0.7394\nEpoch 91/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5325 - accuracy: 0.7401\nEpoch 92/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5329 - accuracy: 0.7402\nEpoch 93/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5328 - accuracy: 0.7407\nEpoch 94/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5324 - accuracy: 0.7404\nEpoch 95/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5317 - accuracy: 0.7410\nEpoch 96/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5321 - accuracy: 0.7408\nEpoch 97/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5325 - accuracy: 0.7405\nEpoch 98/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5323 - accuracy: 0.7408\nEpoch 99/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5325 - accuracy: 0.7415\nEpoch 100/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5320 - accuracy: 0.7408\nEpoch 101/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5323 - accuracy: 0.7402\nEpoch 102/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5318 - accuracy: 0.7409\nEpoch 103/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5317 - accuracy: 0.7411\nEpoch 104/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5329 - accuracy: 0.7411\nEpoch 105/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5314 - accuracy: 0.7411\nEpoch 106/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5310 - accuracy: 0.7411\nEpoch 107/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5316 - accuracy: 0.7407\nEpoch 108/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5320 - accuracy: 0.7407\nEpoch 109/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5319 - accuracy: 0.7409\nEpoch 110/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5315 - accuracy: 0.7401\nEpoch 111/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5316 - accuracy: 0.7403\nEpoch 112/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5314 - accuracy: 0.7405\nEpoch 113/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5315 - accuracy: 0.7413\nEpoch 114/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5315 - accuracy: 0.7406\nEpoch 115/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5318 - accuracy: 0.7399\nEpoch 116/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5321 - accuracy: 0.7406\nEpoch 117/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5319 - accuracy: 0.7416\nEpoch 118/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5320 - accuracy: 0.7402\nEpoch 119/200\n804/804 [==============================] - 3s 4ms/step - loss: 0.5310 - accuracy: 0.7407\nEpoch 120/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5310 - accuracy: 0.7414\nEpoch 121/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5363 - accuracy: 0.7405\nEpoch 122/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5319 - accuracy: 0.7409\nEpoch 123/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5320 - accuracy: 0.7413\nEpoch 124/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5312 - accuracy: 0.7405\nEpoch 125/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5311 - accuracy: 0.7409\nEpoch 126/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5321 - accuracy: 0.7423\nEpoch 127/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5308 - accuracy: 0.7414\nEpoch 128/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5314 - accuracy: 0.7408\nEpoch 129/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5304 - accuracy: 0.7407\nEpoch 130/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5317 - accuracy: 0.7411\nEpoch 131/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5320 - accuracy: 0.7421\nEpoch 132/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5310 - accuracy: 0.7409\nEpoch 133/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5311 - accuracy: 0.7404\nEpoch 134/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5302 - accuracy: 0.7419\nEpoch 135/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5303 - accuracy: 0.7418\nEpoch 136/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5308 - accuracy: 0.7408\nEpoch 137/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5320 - accuracy: 0.7410\nEpoch 138/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5311 - accuracy: 0.7413\nEpoch 139/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5308 - accuracy: 0.7407\nEpoch 140/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7412\nEpoch 141/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5306 - accuracy: 0.7410\nEpoch 142/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5310 - accuracy: 0.7406\nEpoch 143/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5310 - accuracy: 0.7417\nEpoch 144/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7415\nEpoch 145/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7411\nEpoch 146/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7414\nEpoch 147/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5306 - accuracy: 0.7410\nEpoch 148/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5304 - accuracy: 0.7413\nEpoch 149/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5356 - accuracy: 0.7408\nEpoch 150/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5323 - accuracy: 0.7413\nEpoch 151/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5305 - accuracy: 0.7417\nEpoch 152/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5303 - accuracy: 0.7422\nEpoch 153/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5302 - accuracy: 0.7415\nEpoch 154/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5306 - accuracy: 0.7411\nEpoch 155/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7410\nEpoch 156/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5304 - accuracy: 0.7411\nEpoch 157/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5308 - accuracy: 0.7413\nEpoch 158/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7417\nEpoch 159/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5303 - accuracy: 0.7418\nEpoch 160/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5302 - accuracy: 0.7413\nEpoch 161/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5319 - accuracy: 0.7419\nEpoch 162/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5309 - accuracy: 0.7408\nEpoch 163/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5317 - accuracy: 0.7412\nEpoch 164/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5300 - accuracy: 0.7416\nEpoch 165/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5306 - accuracy: 0.7408\nEpoch 166/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5305 - accuracy: 0.7414\nEpoch 167/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5303 - accuracy: 0.7413\nEpoch 168/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5302 - accuracy: 0.7425\nEpoch 169/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5308 - accuracy: 0.7410\nEpoch 170/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5310 - accuracy: 0.7408\nEpoch 171/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5309 - accuracy: 0.7414\nEpoch 172/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5301 - accuracy: 0.7416\nEpoch 173/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5301 - accuracy: 0.7418\nEpoch 174/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5300 - accuracy: 0.7421\nEpoch 175/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5305 - accuracy: 0.7413\nEpoch 176/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5299 - accuracy: 0.7411\nEpoch 177/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5304 - accuracy: 0.7421\nEpoch 178/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5321 - accuracy: 0.7406\nEpoch 179/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5306 - accuracy: 0.7412\nEpoch 180/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5302 - accuracy: 0.7415\nEpoch 181/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5303 - accuracy: 0.7418\nEpoch 182/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5298 - accuracy: 0.7416\nEpoch 183/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5304 - accuracy: 0.7419\nEpoch 184/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5299 - accuracy: 0.7420\nEpoch 185/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5302 - accuracy: 0.7412\nEpoch 186/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5309 - accuracy: 0.7416\nEpoch 187/200\n804/804 [==============================] - 4s 4ms/step - loss: 0.5302 - accuracy: 0.7416\nEpoch 188/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5298 - accuracy: 0.7407\nEpoch 189/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5297 - accuracy: 0.7411\nEpoch 190/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5307 - accuracy: 0.7413\nEpoch 191/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5302 - accuracy: 0.7416\nEpoch 192/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5296 - accuracy: 0.7416\nEpoch 193/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5300 - accuracy: 0.7416\nEpoch 194/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5301 - accuracy: 0.7420\nEpoch 195/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5303 - accuracy: 0.7409\nEpoch 196/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5303 - accuracy: 0.7416\nEpoch 197/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5300 - accuracy: 0.7414\nEpoch 198/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5296 - accuracy: 0.7414\nEpoch 199/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5295 - accuracy: 0.7413\nEpoch 200/200\n804/804 [==============================] - 4s 5ms/step - loss: 0.5300 - accuracy: 0.7407\n" ], [ "# Evaluate the model using the test data\nmodel_loss, model_accuracy = nn4.evaluate(X_test_scaled,y_test,verbose=2)\nprint(f\"Loss: {model_loss}, Accuracy: {model_accuracy}\")", "268/268 - 1s - loss: 0.6118 - accuracy: 0.7266 - 769ms/epoch - 3ms/step\nLoss: 0.6118495464324951, Accuracy: 0.7266472578048706\n" ], [ "# Export our model to HDF5 file\nnn4.save(\"AlphabetSoupCharity_optimize_3.h5\") ", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Numpy", "_____no_output_____" ], [ "We have seen python basic data structures in our last section. They are great but lack specialized features for data analysis. Like, adding rows, columns, operating on 2d matrices aren't readily available. So, we will use *numpy* for such functions.\n\n", "_____no_output_____" ] ], [ [ "import numpy as np", "_____no_output_____" ] ], [ [ "Numpy operates on *nd* arrays. These are similar to lists but contains homogenous elements but easier to store 2-d data.", "_____no_output_____" ] ], [ [ "l1 = [1,2,3,4]\nnd1 = np.array(l1)\nprint(nd1)\n\nl2 = [5,6,7,8]\nnd2 = np.array([l1,l2])\nprint(nd2)", "[1 2 3 4]\n[[1 2 3 4]\n [5 6 7 8]]\n" ] ], [ [ "Sum functions on np.array()", "_____no_output_____" ] ], [ [ "print(nd2.shape)\n\nprint(nd2.size)\n\nprint(nd2.dtype)", "(2, 4)\n8\nint64\n" ] ], [ [ "### Question 1\n\nCreate an identity 2d-array or matrix (with ones across the diagonal).\n\n[ **Hint:** You can also use **np.identity()** function ]", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Question 2\n\nCreate a 2d-array or matrix of order 3x3 with values = 9,8,7,6,5,4,3,2,1 arranged in the same order.\n\nUse: **np.array()** function\n\n", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Question 3\n\nInterchange both the rows and columns of the given matrix.\n\nHint: You can use the transpose **.T**)", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Question 4\nAdd + 1 to all the elements in the given matrix.", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "Similarly you can do operations like scalar substraction, division, multiplication (operating on each element in the matrix)", "_____no_output_____" ], [ "### Question 5\n\nFind the mean of all elements in the given matrix nd6.\nnd6 = [[ 1 4 9 121 144 169]\n [ 16 25 36 196 225 256]\n [ 49 64 81 289 324 361]]\n \n Use: **.mean()** function\n", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Question 7\n\nFind the dot product of two given matrices.\n\n[**Hint:** Use **np.dot()**]", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Array Slicing/Indexing:\n\n- Now we'll learn to access multiple elements or a range of elements from an array.", "_____no_output_____" ] ], [ [ "x = np.arange(20)\nx", "_____no_output_____" ] ], [ [ "### Question 8\n\nPrint the array elements from start to 4th position\n", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Question 9\n\n- Return elements from first position with step size 2. (Difference between consecutive elements is 2)", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "### Question 10\n\nReverse the array using array indexing:", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ], [ "", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%load_ext autoreload\n%autoreload 2\nimport yaml\nimport time\nfrom pathlib import Path", "_____no_output_____" ], [ "import matplotlib as mpl\nimport seaborn as sns\nimport numpy as np\nplt = mpl.pyplot", "_____no_output_____" ], [ "FPS = 30", "_____no_output_____" ], [ "main_cfg_path = \"/home/bo/tmp/presentation-project-attempt4/main_cfg.yaml\"\nwith open(main_cfg_path, \"r\") as stream:\n try:\n main_cfg = yaml.safe_load(stream)\n except yaml.YAMLError as exc:\n raise exc", "_____no_output_____" ], [ "epxt_names = main_cfg[\"experiment_data_paths\"].keys()\nprint(epxt_names)", "dict_keys(['FlyF1-03082020164520', 'FlyF14-08172021175459', 'FlyF15-08182021173222', 'FlyF17-08192021173437', 'FlyF2-03092020175006', 'FlyF20-08202021171828', 'FlyF3-08092021173846', 'FlyF5-08102021175053', 'FlyM12-08162021175307', 'FlyM16-08182021173236', 'FlyM19-08202021171805', 'FlyM21-08262021165434', 'FlyM4-03062020153616', 'FlyM5-03072020150452', 'FlyM8-08112021174107', 'FlyM9-08122021171549'])\n" ], [ "cluster_labels_dict = {}\nfor expt_name in epxt_names:\n cluster_labels_dict[expt_name] = np.load(Path(main_cfg[\"project_path\"]) / expt_name / \"cluster_labels.npy\")", "_____no_output_____" ], [ "# Cluster you're interested in.\ncluster_lbls = [1,2,23,10]\nlbl_name = \"Haltere-switch\"\n\ncluster_occurence_dict = {}\nfor expt_name in epxt_names:\n cluster_labels = cluster_labels_dict[expt_name]\n mask_lbl = np.full(cluster_labels.shape[0], False)\n for lbl in cluster_lbls:\n mask_lbl[cluster_labels == lbl] = True\n print(expt_name,len(np.nonzero(mask_lbl)[0]))\n cluster_occurence_dict[expt_name] = np.nonzero(mask_lbl)[0]", "FlyF1-03082020164520 2856\nFlyF14-08172021175459 1121\nFlyF15-08182021173222 4714\nFlyF17-08192021173437 2877\nFlyF2-03092020175006 2326\nFlyF20-08202021171828 2639\nFlyF3-08092021173846 2259\nFlyF5-08102021175053 3564\nFlyM12-08162021175307 7137\nFlyM16-08182021173236 2139\nFlyM19-08202021171805 3517\nFlyM21-08262021165434 1179\nFlyM4-03062020153616 3107\nFlyM5-03072020150452 2002\nFlyM8-08112021174107 8112\nFlyM9-08122021171549 14888\n" ], [ "cluster_occurence_times = np.array(list(cluster_occurence_dict.values()), dtype=object) / FPS\nfly_expt_names = list(cluster_occurence_dict.keys())\nfig, ax = plt.subplots(figsize=(15,10))\nax = sns.boxplot(data=cluster_occurence_times, orient=\"h\", showfliers=False, palette=\"tab20\", ax=ax)\nax.set_yticklabels(fly_expt_names,)\nformatter = mpl.ticker.FuncFormatter(lambda sec, x: time.strftime(\"%H:%M\", time.gmtime(sec)))\nax.xaxis.set_major_formatter(formatter)\nax.set_title(f\"{lbl_name} Occurence Time Boxplot\", fontsize=40)", "_____no_output_____" ] ], [ [ "## ", "_____no_output_____" ] ] ]

[ "code", "markdown" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# The Graph Data Access\n\nIn this notebook, we read in the data that was generated and saved as a csv from the [TheGraphDataSetCreation](TheGraphDataSetCreation.ipynb) notebook. \n\n\nGoals of this notebook are to obtain:\n\n* Signals, states, event and sequences\n* Volatility metrics\n* ID perceived shocks (correlated with announcements)\n* Signal for target price\n* Signal for market price\n* Error plot\n\nAs a starting point for moving to a decision support system.", "_____no_output_____" ] ], [ [ "# import libraries\nimport pandas as pd\nimport seaborn as sns\nimport matplotlib.pyplot as plt\nimport scipy as sp\nfrom statsmodels.distributions.empirical_distribution import ECDF\nimport scipy.stats as stats", "_____no_output_____" ] ], [ [ "## Import data and add additional attributes", "_____no_output_____" ] ], [ [ "graphData = pd.read_csv('saved_results/RaiLiveGraphData.csv')\ndel graphData['Unnamed: 0']", "_____no_output_____" ], [ "graphData.head()", "_____no_output_____" ], [ "graphData.describe()", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='redemptionPriceActual',kind='line',title='redemptionPriceActual')", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='redemptionRateActual',kind='line',title='redemptionRateActual')", "_____no_output_____" ], [ "graphData['error'] = graphData['redemptionPriceActual'] - graphData['marketPriceUsd']\ngraphData['error_integral'] = graphData['error'].cumsum()", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='error',kind='line',title='error')", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='error_integral',kind='line',title='Steady state error')", "_____no_output_____" ] ], [ [ "## Error experimentation\n\n#### Note: not taking into account control period", "_____no_output_____" ] ], [ [ "kp = 2e-7\n#ki = (-kp * error)/(integral_error)\n# computing at each time, what would the value of ki need to be such that the redemption price would be constant\ngraphData['equilibriation_ki'] = (-kp * graphData.error)/graphData.error_integral", "_____no_output_____" ], [ "# todo iterate through labels and append negative\ngraphData['equilibriation_ki'].apply(lambda x: -x).plot(logy = True,title='Actual equilibriation_ki - flipped sign for log plotting')\nplt.hlines(5e-9, 0, 450, linestyles='solid', label='Recommended ki - flipped sign', color='r')\nplt.hlines(-(graphData['equilibriation_ki'].median()), 0, 450, linestyles='solid', label='median actual ki - flipped', color='g')\nlocs,labels = plt.yticks() # Get the current locations and labelsyticks\nnew_locs = []\nfor i in locs:\n new_locs.append('-'+str(i))\nplt.yticks(locs, new_locs)\nplt.legend(loc=\"upper right\")", "_____no_output_____" ], [ "graphData['equilibriation_ki'].median()", "_____no_output_____" ] ], [ [ "### Counterfactual if intergral control rate had been median the whole time", "_____no_output_____" ] ], [ [ "graphData['counterfactual_redemption_rate'] = (kp * graphData['error'] + graphData['equilibriation_ki'].median())/ graphData['error_integral']", "_____no_output_____" ], [ "subsetGraph = graphData.iloc[50:]\nsns.lineplot(data=subsetGraph,x=\"blockNumber\", y=\"counterfactual_redemption_rate\",label='Counterfactual')\nax2 = plt.twinx()\n# let reflexer know this is wrong\nsns.lineplot(data=subsetGraph,x=\"blockNumber\", y=\"redemptionRateActual\",ax=ax2,color='r',label='Actual')\nplt.title('Actual redemption rate vs counterfactual')\nplt.legend(loc=\"upper left\")\n", "_____no_output_____" ] ], [ [ "## Goodness of fit tests\nWhether or not counterfactual is far enough from actual to reject null that they are from the same distributions.", "_____no_output_____" ] ], [ [ "# fit a cdf\necdf = ECDF(subsetGraph.redemptionRateActual.values)\necdf2 = ECDF(subsetGraph.counterfactual_redemption_rate.values)\n\nplt.plot(ecdf.x,ecdf.y,color='r')\nplt.title('redemptionRateActual ECDF')\nplt.show()\n\nplt.plot(ecdf2.x,ecdf2.y,color='b')\nplt.title('counterfactual_redemption_rate ECDF')\nplt.show()\n\nalpha = 0.05\n\nstatistic, p_value = stats.ks_2samp(subsetGraph.redemptionRateActual.values, subsetGraph.counterfactual_redemption_rate.values) # two sided\nif p_value > alpha:\n decision = \"Sample is from the distribution\"\nelif p_value <= alpha:\n decision = \"Sample is not from the distribution\"\n\nprint(p_value)\nprint(decision)", "_____no_output_____" ] ], [ [ "Based on our analysis using the Kolmogorov-Smirnov Goodness-of-Fit Test, the distributions are very different. As can be seen above from their EDCF plots, you can see a different in their distributions, however pay close attention to the x axis and you can see the distribution difference is significant. ", "_____no_output_____" ] ], [ [ "# scatterplot of linear regressoin residuals\nsns.residplot(x='blockNumber', y='redemptionRateActual', data=subsetGraph, label='redemptionRateActual')\nplt.title('redemptionRateActual regression residuals')", "_____no_output_____" ], [ "sns.residplot(x='blockNumber', y='counterfactual_redemption_rate', data=subsetGraph,label='counterfactual_redemption_rate')\nplt.title('counterfactual_redemption_rate regression residuals')", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='globalDebt',kind='line',title='globalDebt')", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='erc20CoinTotalSupply',kind='line',title='erc20CoinTotalSupply')", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='marketPriceEth',kind='line',title='marketPriceEth')", "_____no_output_____" ], [ "graphData.plot(x='blockNumber',y='marketPriceUsd',kind='line',title='marketPriceUsd')", "_____no_output_____" ] ], [ [ "## Conclusion\n\nUsing The Graph, a lot of data about the Rai system can be obtained for analyzing the health of the system. With some data manipulation, these data streams could be intergrated into the Rai cadCAD model to turn it into a true decision support system.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd\nimport numpy as np\nimport pickle\n\nBASEDIR_MIMIC = '/Volumes/MyData/MIMIC_data/mimiciii/1.4'", "_____no_output_____" ], [ "def get_note_events():\n n_rows = 100000\n\n icd9_code = pd.read_csv(f\"{BASEDIR_MIMIC}/DIAGNOSES_ICD.csv\", index_col = None)\n # create the iterator\n noteevents_iterator = pd.read_csv(\n f\"{BASEDIR_MIMIC}/NOTEEVENTS.csv\",\n iterator=True,\n chunksize=n_rows)\n\n events_list = ['Discharge summary',\n 'Echo',\n 'ECG',\n 'Nursing',\n 'Physician ',\n 'Rehab Services',\n 'Case Management ',\n 'Respiratory ',\n 'Nutrition',\n 'General',\n 'Social Work',\n 'Pharmacy',\n 'Consult',\n 'Radiology',\n 'Nursing/other']\n\n # concatenate according to a filter to get our noteevents data\n noteevents = pd.concat(\n [noteevents_chunk[\n np.logical_and(\n noteevents_chunk.CATEGORY.isin(events_list[1:]),\n noteevents_chunk.DESCRIPTION.isin([\"Report\"])\n )\n ]\n for noteevents_chunk in noteevents_iterator])\n\n # drop all nan in column HADM_ID\n noteevents = noteevents.dropna(subset=[\"HADM_ID\"])\n noteevents.HADM_ID = noteevents.HADM_ID.astype(int)\n try:\n assert len(noteevents.drop_duplicates([\"SUBJECT_ID\",\"HADM_ID\"])) == len(noteevents)\n except AssertionError as e:\n print(\"There are duplicates on Primary Key Set\")\n \n noteevents.CHARTDATE = pd.to_datetime(noteevents.CHARTDATE , format = '%Y-%m-%d %H:%M:%S', errors = 'coerce')\n pd.set_option('display.max_colwidth',50)\n noteevents.sort_values([\"SUBJECT_ID\",\"HADM_ID\",\"CHARTDATE\"], inplace =True)\n #noteevents.drop_duplicates([\"SUBJECT_ID\",\"HADM_ID\"], inplace = True)\n\n noteevents.reset_index(drop = True, inplace = True)\n \n top_values = (icd9_code.groupby('ICD9_CODE').\n agg({\"SUBJECT_ID\": \"nunique\"}).\n reset_index().sort_values(['SUBJECT_ID'], ascending = False).ICD9_CODE.tolist()[:15])\n\n # icd9_code = icd9_code[icd9_code.ICD9_CODE.isin(top_values)]\n icd9_code = icd9_code[icd9_code.ICD9_CODE.isin(top_values)]\n \n import re\n import itertools\n\n def clean_text(text):\n return [x for x in list(itertools.chain.from_iterable([t.split(\"<>\") for t in text.replace(\"\\n\",\" \").split(\" \")])) if len(x) > 0]\n\n\n# irrelevant_tags = [\"Admission Date:\", \"Date of Birth:\", \"Service:\", \"Attending:\", \"Facility:\", \"Medications on Admission:\", \"Discharge Medications:\", \"Completed by:\",\n# \"Dictated By:\" , \"Department:\" , \"Provider:\"]\n\n updated_text = [\"<>\".join([\" \".join(re.split(\"\\n\\d|\\n\\s+\",re.sub(\"^(.*?):\",\"\",x).strip()))\n for x in text.split(\"\\n\\n\")]) for text in noteevents.TEXT]\n updated_text = [re.sub(\"(\\[.*?\\])\", \"\", text) for text in updated_text]\n\n updated_text = [\" \".join(clean_text(x)) for x in updated_text]\n noteevents[\"CLEAN_TEXT\"] = updated_text\n \n return noteevents\n\nnoteevents = get_note_events()", "/var/folders/6p/bxnlvss50bs19f3950g4dfw00000gn/T/ipykernel_36653/2260032821.py:78: DtypeWarning: Columns (5) have mixed types.Specify dtype option on import or set low_memory=False.\n noteevents = get_note_events()\n" ], [ "def mapNotes(dataset):\n print(f\"Mapping notes on {dataset}.\")\n df = pickle.load(open(f'../data/mimic3/train_data_mimic3/{dataset}', 'rb'))\n \n BASEDIR_MIMIC = '/Volumes/MyData/MIMIC_data/mimiciii/1.4'\n icustays = pd.read_csv(f\"{BASEDIR_MIMIC}/ICUSTAYS.csv\", index_col = None)\n \n # SUBJECT_ID \"_\" ICUSTAY_ID \"_episode\" episode \"_timeseries_readmission.csv\"\n\n import re\n \n episodes = df['names']\n \n regex = r\"(\\d+)_(\\d+)_episode(\\d+)_timeseries_readmission\\.csv\"\n\n sid = []\n hadmids = []\n icustayid = [] # ICUSTAYS.csv ICUSTAY_ID\n episode = []\n notestexts = []\n notextepis = []\n for epi in episodes:\n match = re.findall(regex, epi) #, re.MULTILINE)\n sid.append(int(match[0][0]))\n icustayid.append(int(match[0][1]))\n episode.append(int(match[0][2]))\n hadmid = icustays[icustays['ICUSTAY_ID']==int(match[0][1])]['HADM_ID']\n hadmids.append(int(hadmid))\n try:\n #text = noteevents[noteevents['HADM_ID']==int(hadmid)]['TEXT'].iloc[0]\n #text = noteevents[noteevents['HADM_ID']==int(hadmid)]['CLEAN_TEXT'].iloc[0]\n text = \"\\n\\n\".join([t for t in noteevents[noteevents['HADM_ID']==int(hadmid)]['CLEAN_TEXT']])\n except:\n notextepis.append(int(hadmid))\n text = ''\n notestexts.append(text)\n\n print(len(episodes), len(notextepis), len(set(notextepis)))\n print(len(sid), len(hadmids), len(df['names']))\n \n notesfull = pd.DataFrame({'SUBJECT_ID':sid, 'HADM_ID':hadmids, 'ICUSTAY_ID':icustayid, 'EPISODE':episode, 'CLEAN_TEXT':notestexts})\n \n # save full data\n filename = f'./events_notes_{dataset}'\n\n with open(filename + '.pickle', 'wb') as handle:\n pickle.dump(notesfull, handle, protocol=pickle.HIGHEST_PROTOCOL)\n \n print(f\"Finished mapping notes on {dataset}.\\n\")", "_____no_output_____" ], [ "def combineData(dataset):\n print(f\"Combining data for all {dataset}.\")\n df = pickle.load(open(f'../data/mimic3/train_data_mimic3/{dataset}', 'rb'))\n print(df.keys(), len(df['data']),len(df['names']), df['data'][0].shape, len(df['data'][1]), len(df['names']))\n\n notes = pickle.load(open(f'clinical_notes_{dataset}.pickle', 'rb'))\n eventsnotes = pickle.load(open(f'events_notes_{dataset}.pickle', 'rb'))\n\n # how many empty text rows\n # np.where(notes.applymap(lambda x: x == ''))\n\n # how many empty text rows\n print(f\"There are {len(list(notes[notes['CLEAN_TEXT'] == ''].index))} empty rows in notes.\")\n print(f\"There are {len(list(eventsnotes[eventsnotes['CLEAN_TEXT'] == ''].index))} empty rows in eventsnotes.\")\n \n X = df['data'][0]\n y = np.array(df['data'][1])\n N = list(notes.CLEAN_TEXT)\n EN = list(eventsnotes.CLEAN_TEXT)\n\n # check if all three data sets have the same size/length\n assert len(X) == len(y) == len(N) == len(EN)\n\n empty_ind_N = list(notes[notes['CLEAN_TEXT'] == ''].index)\n empty_ind_EN = list(notes[eventsnotes['CLEAN_TEXT'] == ''].index)\n N_ = np.array(N)\n EN_ = np.array(EN)\n\n mask = np.ones(len(notes), np.bool)\n mask[empty_ind_N] = 0\n mask[empty_ind_EN] = 0\n \n good_notes = N_[mask]\n good_eventsnotes = EN_[mask]\n good_X = X[mask]\n good_y = y[mask]\n\n print(f\"Final shapes = {good_X.shape, good_y.shape, good_notes.shape}\")\n\n data = {'inputs': good_X,\n 'labels': good_y,\n 'eventsnotes': good_eventsnotes,\n 'notes': good_notes}\n\n # save full data\n filename = f'./new_{dataset}_CNEP'\n #full_data.to_csv(filename + '.csv', index = None)\n\n with open(filename + '.pickle', 'wb') as handle:\n pickle.dump(data, handle, protocol=pickle.HIGHEST_PROTOCOL)\n \n print(\"finished.\\n\")", "_____no_output_____" ], [ "all_datasets = ['train_data', 'test_data', 'val_data']\n\nfor dataset in all_datasets:\n print(f\"\\n\\nProcessing dataset {dataset}.\")\n mapNotes(dataset)\n combineData(dataset)\n ", "\n\nProcessing dataset train_data.\nMapping notes on train_data.\n15392 0 0\n15392 15392 15392\nFinished mapping notes on train_data.\n\nCombining data for all train_data.\ndict_keys(['data', 'names']) 2 15392 (15392, 48, 390) 15392 15392\nThere are 374 empty rows in notes.\nThere are 680 empty rows in eventsnotes.\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]