Box Plot Statistics

box_describe.py

def box_describe(df_copy):
    df_stats = df_copy.describe().T
    df_stats.rename(columns = {
        '25%': 'q1',
        '50%': 'q2',
        '75%': 'q3'
    }, inplace = True)
    
    df_stats['iqr'] = df_stats['q3'] - df_stats['q1']
    df_stats['lower_fence'] = df_stats['q1'] - (1.5 * df_stats['iqr'])  # Lower Fence
    df_stats['upper_fence'] = df_stats['q3'] + (1.5 * df_stats['iqr'])  # Upper Fence
    
    def get_outliers(series):
        lower_outliers = (series < df_stats.loc[series.name].lower_fence).sum()
        upper_outliers = (series > df_stats.loc[series.name].upper_fence).sum()
        return lower_outliers + upper_outliers
    
    df_stats['outliers'] = df_copy.apply(lambda x: get_outliers(x))
    df_stats['outlier%'] = (df_stats['outliers'] / df_stats['count']) * 100
    
    return df_stats