3. Grouped Summary Statistics

Summaries by groups

dogs[dogs[“color”] == “Black”][“weight_kg”].mean()

dogs[dogs[“color”] == “Brown”][“weight_kg”].mean()

dogs[dogs[“color”] == “White”][“weight_kg”].mean()

dogs[dogs[“color”] == “Gray”][“weight_kg”].mean()

Grouped summary statistics

dogs.groupby(“color”)[“weight_kg”].mean()

Multiple grouped summaries

• Using aggregate method to get multiple statistics

dogs.groupby(“color”)[“weight_kg”].agg([min, max, sum])

Grouping by multiple variables

dogs.groupby([“color”, “breed”])[“weight_kg”].mean()

Many groups, many summaries

dogs.groupby([“color”, “breed”])[[“weight_kg”, “height_cm”]].mean()

Example 1

sales_by_type = sales.groupby("type")["weekly_sales"].sum() # Groupby type and calculate total weekly sales

sales_propn_by_type = sales_by_type / sales["weekly_sales"].sum() # Get proportion for each type

print(sales_propn_by_type)

sales_by_type_is_holiday = sales.groupby(["type","is_holiday"])["weekly_sales"].sum() # Groupby type and is_holiday, and calculate total weekly sales

print(sales_by_type_is_holiday)

Example 2

import numpy as np

sales_stats = sales.groupby("type")["weekly_sales"].agg([np.min, np.max, np.mean, np.median]) # For each store type, aggregate weekly_sales

print(sales_stats)

unemp_fuel_stats = sales.groupby("type")[["unemployment","fuel_price_usd_per_l"]].agg([np.min, np.max, np.mean, np.median]) # For each store type, aggregate unemployment and fuel_price_usd_per_l

print(unemp_fuel_stats)