transform versus aggregate

One answer per group, or one answer per row.

aggregate collapses

A group of five rows becomes one. That is what you want for a summary table and not what you want for a new column.

import pandas as pd

df = pd.DataFrame({"k": ["a", "a", "b"], "v": [1, 3, 5]})
print(df.groupby("k")["v"].mean().to_dict())

transform keeps the shape

It gives every row its group's answer, so it lines up with the table and can be assigned straight back as a column.

import pandas as pd

df = pd.DataFrame({"k": ["a", "a", "b"], "v": [1, 3, 5]})
df["group_mean"] = df.groupby("k")["v"].transform("mean")
print(df["group_mean"].tolist())

Exercise

Try It Yourself

Write above_group_mean(df) returning the number of rows whose v is above the mean of their own group.

import pandas as pd

def above_group_mean(df):
    # Rows beating their own group average.
    return 0

print(above_group_mean(pd.DataFrame({"k": ["a", "a"], "v": [1, 3]})))

Press Run to see output

Check Your Understanding

← Previous Next