easy-RAPM.py

import numpy as np
from sklearn import linear_model
import pandas as pd
import requests
from sklearn.feature_extraction import DictVectorizer

team_id_list = ['1610612737','1610612738','1610612751','1610612766','1610612741','1610612739','1610612742','1610612743','1610612765','1610612744','1610612745','1610612754','1610612746','1610612747','1610612763','1610612748','1610612749','1610612750','1610612740','1610612752','1610612760','1610612753','1610612755','1610612756','1610612757','1610612758','1610612759','1610612761','1610612762','1610612764']


headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/61.0.3163.100 Safari/537.36'}

df = pd.DataFrame()
for team_id in team_id_list: 
    params = (
        ('Conference', ''),
        ('DateFrom', ''),
        ('DateTo', ''),
        ('Division', ''),
        ('GameID', ''),
        ('GameSegment', ''),
        ('GroupQuantity', '5'),
        ('LastNGames', '0'),
        ('LeagueID', '00'),
        ('Location', ''),
        ('MeasureType', 'Advanced'),
        ('Month', '0'),
        ('OpponentTeamID', '0'),
        ('Outcome', ''),
        ('PORound', '0'),
        ('PaceAdjust', 'N'),
        ('PerMode', 'PerGame'),
        ('Period', '0'),
        ('PlusMinus', 'N'),
        ('Rank', 'N'),
        ('Season', '2017-18'),
        ('SeasonSegment', ''),
        ('SeasonType', 'Regular Season'),
        ('ShotClockRange', ''),
        ('TeamID', team_id),
        ('VsConference', ''),
        ('VsDivision', ''),
    )

    data = requests.get('https://stats.nba.com/stats/leaguedashlineups', headers=headers, params=params).json()
    df = df.append(pd.DataFrame(data['resultSets'][0]['rowSet']))

df.columns = data['resultSets'][0]['headers']

units = []
ORTG = []
DRTG = []
weights = []
for i, name in enumerate(df['GROUP_NAME']):

    temp_string = name.split(' - ')
    
    home_offense_unit = {name :1 for name in temp_string}
    units.append(home_offense_unit)
    ORTG.append(df['OFF_RATING'].iloc[i])
    DRTG.append(df['DEF_RATING'].iloc[i])
    weights.append(df['MIN'].iloc[i])
    

u = DictVectorizer(sparse=False)
u_mat = u.fit_transform(units)
players = u.get_feature_names()


clf = linear_model.RidgeCV(alphas=(np.array([3000])), cv=5)
weights = np.asarray(weights)
clf.fit(u_mat, ORTG, sample_weight=weights)
off_ratings = []
for player in players:
    off_ratings.append((player, clf.coef_[players.index(player)]))
    
clf.fit(u_mat, DRTG, sample_weight=weights)    
def_ratings = []
for player in players:
    def_ratings.append((player, clf.coef_[players.index(player)]))
    
player_name_list = []
ORAPM_list = []
DRAPM_list = []
for rating in off_ratings:

    player_name_list.append(rating[0])
    ORAPM_list.append(rating[1])
for rating in def_ratings:
    DRAPM_list.append(-rating[1])
    
RAPM_list = []
for i in range(len(ORAPM_list)):
    RAPM_list.append(ORAPM_list[i]+DRAPM_list[i]) 
RAPM_dict = {'Player': player_name_list, 'ORAPM': ORAPM_list,'DRAPM': DRAPM_list,'RAPM': RAPM_list }
RAPM_df = pd.DataFrame(data=RAPM_dict)
RAPM_df = RAPM_df.sort_values(by=['RAPM'],ascending=False) 
print(RAPM_df)