Grouping and Merging

Two operations unlock most data analysis: groupby (split rows into groups and summarise each) and merge (join two tables on a shared key).

groupby follows a split-apply-combine pattern. Group by a column, then aggregate:

import pandas as pd
sales = pd.DataFrame({
    "region": ["N", "S", "N", "S"],
    "amount": [100, 200, 150, 50],
})
print(sales.groupby("region")["amount"].sum())
# region
# N    250
# S    250

Use .agg to compute several summaries at once:

print(sales.groupby("region")["amount"].agg(["sum", "mean", "count"]))

merge joins tables that share a column. Here both tables have region:

info = pd.DataFrame({
    "region":  ["N", "S"],
    "manager": ["Ada", "Bo"],
})
joined = sales.merge(info, on="region")
print(joined)   # each sales row gains its region's manager

The default is an inner join: only keys present in both tables survive. Pass how="left" to keep every row of the left table even when there is no match.