[pandas - group small values] #python #pandas #data #datamanipulation #dataframe #jupyter #notebook

group_small_values.py

def group_small_values(data, idx, column, percentage):
    """ 
    Groups small dataframe values into a sigle group
        - data = source dataframe
        - idx = column with qualitative values
        - column = column to evaluate
        - percentage = group values which value > sum(column)
    """
    total_column = data[column].sum()
    other_percentage = ((float(total_column) / float(100) * float(percentage)))
    data.loc[data[column] <other_percentage, 'group'] = '> ' + percentage
    data.loc[data[column] >=other_percentage, 'group'] = data[idx]
    return(data.groupby('group', as_index=False).agg({column: "sum"}).sort_values([column]))