Spaces:
Running
Running
File size: 15,190 Bytes
dfb2f96 1a2fb86 dfb2f96 1a2fb86 dfb2f96 1a2fb86 0573513 1a2fb86 0573513 1a2fb86 a11d086 0e788d9 6a86553 ed2133a 1efb29a 4ea6358 d05f3c9 86629c5 a11d086 dfb2f96 1a2fb86 a11d086 dfb2f96 ae85aee a11d086 aa723b8 a11d086 5586c8f aa723b8 5586c8f aa723b8 5586c8f 29ed0ed 5586c8f 876e35b 5586c8f 1e2f35d 659ff69 5586c8f 1e2f35d 5586c8f d317832 5586c8f 1e2f35d 5586c8f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 |
import pulp
import numpy as np
import pandas as pd
import streamlit as st
import gspread
import time
import random
import scipy.stats
@st.cache_resource
def init_conn():
scope = ['https://www.googleapis.com/auth/spreadsheets',
"https://www.googleapis.com/auth/drive"]
credentials = {
"type": "service_account",
"project_id": "sheets-api-connect-378620",
"private_key_id": "1005124050c80d085e2c5b344345715978dd9cc9",
"private_key": "-----BEGIN PRIVATE KEY-----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\nJuElxA33rFEjbgbzdyrE768=\n-----END PRIVATE KEY-----\n",
"client_email": "gspread-connection@sheets-api-connect-378620.iam.gserviceaccount.com",
"client_id": "106625872877651920064",
"auth_uri": "https://accounts.google.com/o/oauth2/auth",
"token_uri": "https://oauth2.googleapis.com/token",
"auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
"client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/gspread-connection%40sheets-api-connect-378620.iam.gserviceaccount.com"
}
gc = gspread.service_account_from_dict(credentials)
return gc
st.set_page_config(layout="wide")
gc = init_conn()
game_format = {'Dropback% Proj': '{:.2%}', 'DesRush%': '{:.2%}', 'Rush%': '{:.2%}'}
rb_util = {'Player Snaps%': '{:.2%}','Rush Att%': '{:.2%}', 'Routes%': '{:.2%}', 'Targets%': '{:.2%}', 'SDD Snaps%': '{:.2%}', 'i5 Rush%': '{:.2%}',
'LDD Snaps%': '{:.2%}','2-min%': '{:.2%}'}
wr_te_util = {'Routes%': '{:.2%}','Targets%': '{:.2%}', 'Air Yards%': '{:.2%}', 'Endzone Targets%': '{:.2%}', 'Third/Fourth%': '{:.2%}', 'Third/Fourth Targets%': '{:.2%}',
'Play Action Targets%': '{:.2%}','2-min%': '{:.2%}'}
all_dk_player_projections = 'https://docs.google.com/spreadsheets/d/1I_1Ve3F4tftgfLQQoRKOJ351XfEG48s36OxXUKxmgS8/edit#gid=179416653'
@st.cache_resource(ttl = 300)
def rb_util_weekly():
sh = gc.open_by_url(all_dk_player_projections)
worksheet = sh.worksheet('RB_Util')
raw_display = pd.DataFrame(worksheet.get_all_records())
raw_display = raw_display.replace('', np.nan)
raw_display = raw_display[['player_name', 'position', 'week', 'team_season', 'player_snaps_per', 'rush_attempts_per', 'routes_per', 'targets_per',
'tprr', 'player_SDD_snaps_per', 'inside_five_rush_per', 'player_LDD_snaps_per', 'two_min_per', 'exPPR', 'ppr_fantasy', 'UR_Rank']]
raw_display = raw_display.set_axis(['Player', 'Position', 'Week', 'Team-Season', 'Player Snaps%', 'Rush Att%', 'Routes%', 'Targets%',
'TPRR', 'SDD Snaps%', 'i5 Rush%', 'LDD Snaps%', '2-min%', 'Expected PPR', 'PPR', 'Utilization Rank'], axis='columns')
raw_display = raw_display.sort_values(by='Utilization Rank', ascending=True)
return raw_display
@st.cache_resource(ttl = 300)
def wr_te_util_weekly():
sh = gc.open_by_url(all_dk_player_projections)
worksheet = sh.worksheet('WR_TE_Util')
raw_display = pd.DataFrame(worksheet.get_all_records())
raw_display = raw_display.replace('', np.nan)
raw_display = raw_display[['player_name', 'position', 'week', 'team_season', 'routes_per', 'targets_per', 'tprr' , 'adot', 'air_yards_per',
'ayprr', 'endzone_targets_per', 'third_fourth_per', 'third_fourth_target_per', 'play_action_targets_per', 'exPPR', 'ppr_fantasy', 'UR_Rank']]
raw_display = raw_display.set_axis(['Player', 'Position', 'Week', 'Team-Season', 'Routes%', 'Targets%', 'TPRR' , 'ADOT', 'Air Yards%',
'AYPRR', 'Endzone Targets%', 'Third/Fourth%', 'Third/Fourth Targets%', 'Play Action Targets%', 'Expected PPR', 'PPR', 'Utilization Rank'], axis='columns')
raw_display = raw_display.sort_values(by='Utilization Rank', ascending=True)
return raw_display
@st.cache_resource(ttl = 300)
def rb_util_season():
sh = gc.open_by_url(all_dk_player_projections)
worksheet = sh.worksheet('RB_Util_Season')
raw_display = pd.DataFrame(worksheet.get_all_records())
raw_display = raw_display.replace('', np.nan)
raw_display = raw_display[['player_name', 'position', 'team_season', 'player_snaps_per', 'rush_attempts_per', 'routes_per', 'targets_per',
'tprr', 'player_SDD_snaps_per', 'inside_five_rush_per', 'player_LDD_snaps_per', 'two_min_per', 'exPPR', 'ppr_fantasy', 'UR_Rank']]
raw_display = raw_display.set_axis(['Player', 'Position', 'Team-Season', 'Player Snaps%', 'Rush Att%', 'Routes%', 'Targets%',
'TPRR', 'SDD Snaps%', 'i5 Rush%', 'LDD Snaps%', '2-min%', 'Expected PPR', 'PPR', 'Utilization Rank'], axis='columns')
raw_display = raw_display.sort_values(by='Utilization Rank', ascending=True)
return raw_display
@st.cache_resource(ttl = 300)
def wr_te_util_season():
sh = gc.open_by_url(all_dk_player_projections)
worksheet = sh.worksheet('WR_TE_Util_Season')
raw_display = pd.DataFrame(worksheet.get_all_records())
raw_display = raw_display.replace('', np.nan)
raw_display = raw_display[['player_name', 'position', 'team_season', 'routes_per', 'targets_per', 'tprr' , 'adot', 'air_yards_per',
'ayprr', 'endzone_targets_per', 'third_fourth_per', 'third_fourth_target_per', 'play_action_targets_per', 'exPPR', 'ppr_fantasy', 'UR_Rank']]
raw_display = raw_display.set_axis(['Player', 'Position', 'Team-Season', 'Routes%', 'Targets%', 'TPRR' , 'ADOT', 'Air Yards%',
'AYPRR', 'Endzone Targets%', 'Third/Fourth%', 'Third/Fourth Targets%', 'Play Action Targets%', 'Expected PPR', 'PPR', 'Utilization Rank'], axis='columns')
raw_display = raw_display.sort_values(by='Utilization Rank', ascending=True)
return raw_display
@st.cache_resource(ttl = 300)
def coverage_matchups():
sh = gc.open_by_url(all_dk_player_projections)
worksheet = sh.worksheet('Defensive Matchups')
raw_display = pd.DataFrame(worksheet.get_all_records())
raw_display = raw_display.replace('', np.nan)
raw_display = raw_display.dropna(subset='Weighted Targets')
raw_display = raw_display[raw_display['Weighted Targets'] != '#DIV/0!']
raw_display = raw_display[raw_display['Weighted Targets'] != '#N/A']
raw_display = raw_display.apply(pd.to_numeric(errors='ignore'))
# raw_display = raw_display.sort_values(by='Weighted Targets', ascending=False)
return raw_display
@st.cache_resource(ttl = 300)
def macro_pull():
sh = gc.open_by_url(all_dk_player_projections)
worksheet = sh.worksheet('FL_Macro')
raw_display = pd.DataFrame(worksheet.get_all_records())
raw_display = raw_display.sort_values(by='Team Total', ascending=False)
return raw_display
@st.cache_data
def convert_df_to_csv(df):
return df.to_csv().encode('utf-8')
rb_search = rb_util_weekly()
wr_search = wr_te_util_weekly()
rb_season = rb_util_season()
wr_season = wr_te_util_season()
wr_matchups = coverage_matchups()
macro_data = macro_pull()
pos_list = ['RB', 'WR', 'TE']
tab1, tab2 = st.tabs(["Season Long Research", "Slate Specific"])
with tab1:
col1, col2 = st.columns([1, 8])
with col1:
if st.button("Load/Reset Data", key='reset1'):
st.cache_data.clear()
rb_search = rb_util_season()
wr_search = wr_te_util_season()
rb_season = rb_util_season()
wr_season = wr_te_util_season()
macro_data = macro_pull()
stat_type_var1 = st.radio("What table are you loading?", ('Macro Table', 'RB Usage (Weekly)', 'WR/TE Usage (Weekly)', 'RB Usage (Season)', 'WR/TE Usage (Season)'), key='stat_type_var1')
split_var1 = st.radio("Are you running the the whole league or certain teams?", ('All Teams', 'Specific Teams'), key='split_var1')
pos_split1 = st.radio("Are you viewing all positions or specific positions?", ('All Positions', 'Specific Positions'), key='pos_split1')
if pos_split1 == 'Specific Positions':
pos_var1 = st.multiselect('What Positions would you like to view?', options = ['RB', 'WR', 'TE'])
elif pos_split1 == 'All Positions':
pos_var1 = pos_list
if split_var1 == 'Specific Teams':
team_var1 = st.multiselect('Which teams would you like to include in the Table?', options = rb_search['Team-Season'].unique(), key='team_var1')
elif split_var1 == 'All Teams':
team_var1 = rb_search['Team-Season'].unique().tolist()
if stat_type_var1 == 'Macro Table':
table_instance = macro_data
table_instance = table_instance.set_index('team')
elif stat_type_var1 == 'RB Usage (Weekly)':
table_instance = rb_search
table_instance = table_instance[table_instance['Team-Season'].isin(team_var1)]
table_instance = table_instance[table_instance['Position'].isin(pos_var1)]
elif stat_type_var1 == 'WR/TE Usage (Weekly)':
table_instance = wr_search
table_instance = table_instance[table_instance['Team-Season'].isin(team_var1)]
table_instance = table_instance[table_instance['Position'].isin(pos_var1)]
elif stat_type_var1 == 'RB Usage (Season)':
table_instance = rb_season
table_instance = table_instance[table_instance['Team-Season'].isin(team_var1)]
table_instance = table_instance[table_instance['Position'].isin(pos_var1)]
elif stat_type_var1 == 'WR/TE Usage (Season)':
table_instance = wr_season
table_instance = table_instance[table_instance['Team-Season'].isin(team_var1)]
table_instance = table_instance[table_instance['Position'].isin(pos_var1)]
with col2:
if stat_type_var1 == 'Macro Table':
st.dataframe(table_instance.style.background_gradient(axis=0).background_gradient(cmap = 'RdYlGn').format(game_format, precision=2), use_container_width = True)
elif stat_type_var1 == 'RB Usage (Weekly)':
st.dataframe(table_instance.style.background_gradient(axis=0).background_gradient(cmap = 'RdYlGn').background_gradient(cmap='RdYlGn_r', subset = 'Utilization Rank').format(rb_util, precision=2), use_container_width = True)
elif stat_type_var1 == 'WR/TE Usage (Weekly)':
st.dataframe(table_instance.style.background_gradient(axis=0).background_gradient(cmap = 'RdYlGn').background_gradient(cmap='RdYlGn_r', subset = 'Utilization Rank').format(wr_te_util, precision=2), use_container_width = True)
elif stat_type_var1 == 'RB Usage (Season)':
st.dataframe(table_instance.style.background_gradient(axis=0).background_gradient(cmap = 'RdYlGn').background_gradient(cmap='RdYlGn_r', subset = 'Utilization Rank').format(rb_util, precision=2), use_container_width = True)
elif stat_type_var1 == 'WR/TE Usage (Season)':
st.dataframe(table_instance.style.background_gradient(axis=0).background_gradient(cmap = 'RdYlGn').background_gradient(cmap='RdYlGn_r', subset = 'Utilization Rank').format(wr_te_util, precision=2), use_container_width = True)
st.download_button(
label="Export Tables",
data=convert_df_to_csv(table_instance),
file_name='NFL_Research_export.csv',
mime='text/csv',
)
with tab2:
col1, col2 = st.columns([1, 8])
with col1:
if st.button("Load/Reset Data", key='reset2'):
st.cache_data.clear()
wr_matchups = coverage_matchups()
stat_type_var2 = st.radio("What table are you loading?", ('WR/TE Coverage Matchups', 'Nothing idk lol'))
split_var2 = st.radio("Are you running the the whole league or certain teams?", ('All Teams', 'Specific Teams'))
pos_split2 = st.radio("Are you viewing all positions or specific positions?", ('All Positions', 'Specific Positions'))
if pos_split2 == 'Specific Positions':
pos_var2 = st.multiselect('What Positions would you like to view?', options = ['RB', 'WR', 'TE'])
elif pos_split2 == 'All Positions':
pos_var2 = pos_list
if split_var2 == 'Specific Teams':
team_var2 = st.multiselect('Which teams would you like to include in the Table?', options = wr_matchups['Team'].unique())
elif split_var2 == 'All Teams':
team_var2 = wr_matchups['Team'].unique().tolist()
if stat_type_var2 == 'WR/TE Coverage Matchups':
slate_table_instance = wr_matchups
slate_table_instance = slate_table_instance.set_index('name')
elif stat_type_var1 == 'Nothing idk lol':
slate_table_instance = wr_matchups
with col2:
st.table(wr_matchups)
if stat_type_var2 == 'WR/TE Coverage Matchups':
st.dataframe(slate_table_instance.style.background_gradient(axis=0).background_gradient(cmap = 'RdYlGn').format(precision=2), use_container_width = True)
elif stat_type_var2 == 'Nothing idk lol':
st.write('lol same bro but yo the vibes immaculate')
st.download_button(
label="Export Tables",
data=convert_df_to_csv(slate_table_instance),
file_name='NFL_Slate_Research_export.csv',
mime='text/csv',
) |