In [1]:

    

import pandas as pd
import numpy as np


    

In [6]:

    

def off_win(row, games_df):
    score = games_df.iloc[row.loc['gid'] - 1]

    if row['off'] == score['h']:

        if score.ptsh > score.ptsv:
            return 1
        else:
            return 0


    elif row['off'] == score['v']:

        if score.ptsv > score.ptsh:
            return 1

        else:
            return 0

    else:
        raise Exception('Teams do not match: %s vs. %s; %s vs. %s' % (
            row['off'], row['def'], score['v'], score['h']))


    

In [7]:

    

plays_df = pd.read_csv('../data/raw/PLAY.csv')
games_df = pd.read_csv('../data/raw/GAME.csv')


    

In [8]:

    

plays_df = plays_df.loc[:, [
    'gid',
    'off',
    'def',
    'qtr',
    'min',
    'sec',
    'ptso',
    'ptsd',
    'timo',
    'timd',
    'dwn',
    'ytg',
    'yfog'
]]


    

In [10]:

    

plays_df['y'] = plays_df.apply(lambda row: off_win(row, games_df), axis=1)
plays_df.to_csv('../data/Xy.csv')


    

In [ ]: