load_expanded_shooting_df.py

def load_expanded_shooting_df():
    shot_df = load_shooting_df()

    # Load Team Info
    team_info = pd.read_csv('data/team_info.csv')
    team_info['combined_name'] = team_info.shortName + ' ' + team_info.teamName
    team_info.drop(
        ['franchiseId', 'shortName', 'teamName', 'abbreviation', 'link'],
        axis=1,
        inplace=True)

    # Load Games Dataset
    games_df = pd.read_csv('data/game.csv')
    games_df.drop(['venue_link', 'venue_time_zone_id'], axis=1, inplace=True)

    # Load and Drop Unused Columns
    team_info = pd.read_csv('data/team_info.csv')
    team_info['combined_name'] = team_info.shortName + ' ' + team_info.teamName
    team_info.drop(
        ['franchiseId', 'shortName', 'teamName', 'abbreviation', 'link'],
        axis=1,
        inplace=True)

    # Create home and away datasets for joining
    away_info = team_info.copy()
    away_info.columns = ['away_team_id', 'away_team_name']
    home_info = team_info.copy()
    home_info.columns = ['home_team_id', 'home_team_name']

    # Merge Columns
    games_df = games_df.merge(away_info)
    games_df = games_df.merge(home_info)
    shot_df = shot_df.merge(games_df, on='game_id')

    # Get unique games
    unique_games = shot_df.game_id.unique()
    print("Loaded {} Unique Games".format(len(unique_games)))
    time.sleep(0.5)  # For printing issue in notebooks
    return shot_df, unique_games