2. Aggregating Data

Summary Statistics

• Summarizing numerical data
  • .mean()

print(sales["weekly_sales"].mean())

• .mode()

print(sales["weekly_sales"].mode())

• .median()

print(sales["weekly_sales"].median())

• .min()
• .max()
• .var()
• .std()
• .sum()
• .quantile()

Summarizing dates

• students[“date_of_birth”].min() -> Oldest student
• students[“date_of_birth”].max() -> Youngest student
• Example:

print(sales["date"].max()) # Print the maximum of the date column

print(sales["date"].min()) # Print the minimum of the date column

The .agg() method

• Allows to compute custom summary statistics
• Create a function called pct30 that computes the thirtieth percentile of a DataFrame column

def pct30(column):

return colum.quantile(0.3)

dogs[“weight_kg”].agg(pct30)

dogs[[“weight_kg”, “height_”cm]].agg(pct30) # Summaries on multiple columns

• Multiple summaries

def pct40(column):

return colum.quantile(0.4)

dogs[“weight_kg”].agg([pct30, pct40]) # Multiple summaries

Cumulative statistics

• .cumsum(
  • cumulative sum on a column returns a number for each row of the DataFrame
• .cummax()
  • cumulative maximum on a column returns a number for each row of the DataFrame
• .cummin()
  • cumulative minimum on a column returns a number for each row of the DataFrame
• .cumprod()
  • cumulative product on a column returns a number for each row of the DataFrame

Examples

• Example 1

import numpy as np

def iqr(column): # A custom IQR function

return column.quantile(0.75) - column.quantile(0.25)

print(sales["temperature_c"].agg(iqr)) # Update to print IQR

print(sales[["temperature_c","fuel_price_usd_per_l","unemployment"]].agg(iqr)) # Update to print IQR

print(sales[["temperature_c", "fuel_price_usd_per_l", "unemployment"]].agg([iqr, np.median])) # Update to print IQR and median

• Example 2

sales_1_1 = sales_1_1.sort_values("date", ascending=True) # Sort sales_1_1 by date

sales_1_1["cum_weekly_sales"] = sales_1_1["weekly_sales"].cumsum() # Get the cumulative sum of weekly_sales

sales_1_1["cum_max_sales"] = sales_1_1["weekly_sales"].cummax() # Get the cumulative max of weekly_sales

print(sales_1_1[["date", "weekly_sales", "cum_weekly_sales", "cum_max_sales"]])