seasonal_df.py

def get_seasonal_reb_df():
    shot_to_reb = load_rebounds_df()
    ## Use only regular season
    shot_to_reb = shot_to_reb[shot_to_reb.type == 'R']

    rebound_df = shot_to_reb.groupby(['for_name', 'season', 'lead_to_reb']).agg({
        'counter':
        'count'
    }).reset_index()

    rebound_df = rebound_df.pivot_table(
        index=['season', 'for_name'], columns='lead_to_reb').reset_index()
    rebound_df.columns = rebound_df.columns.droplevel()
    rebound_df.columns = ['season', 'team_name', 'no_rebound', 'rebound']

    rebound_df['total_shots'] = rebound_df['no_rebound'] + rebound_df['rebound']
    rebound_df['rebound_ratio'] = rebound_df['rebound'] / rebound_df['total_shots']

    ## Remove 2012 (only half a season)
    rebound_df = rebound_df[rebound_df.season != '2012']

    ## Get seasonal data and merge
    seasonal_df = get_seasonal_summary('R')
    seasonal_df = rebound_df.merge(
        seasonal_df,
        right_on=['season', 'combined_name'],
        left_on=['season', 'team_name'])

    #Remove season-team pairs with far too low shot count in the play_df
    seasonal_df = seasonal_df[seasonal_df.total_shots > seasonal_df.shots * 0.75]
    return seasonal_df