groupby — the move that pays for the course

The pattern

# Total revenue per region — one line
df.groupby("region")["amount"].sum()

That's it. That replaces a column of SUMIFS formulas, plus the formula maintenance, plus the broken refs when rows shift.

Multiple aggregations

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

df.groupby("region").agg(
    total_revenue = ("amount",   "sum"),
    avg_order     = ("amount",   "mean"),
    n_orders      = ("order_id", "count"),
)

The second form (named aggregations) is the cleanest — output columns are named exactly what you want.

Multiple group keys

df.groupby(["region", "quarter"])["amount"].sum()

Get a clean DataFrame back

By default, groupby returns a Series or a DataFrame with a hierarchical index. Add .reset_index() for a flat one:

summary = (df.groupby(["region", "quarter"])["amount"]
             .sum()
             .reset_index())

Filtering within groups

# Keep only groups with more than 10 orders
df.groupby("region").filter(lambda g: len(g) > 10)

# Top 3 per group
df.sort_values("amount", ascending=False).groupby("region").head(3)

Walkthrough: the classic regional summary

df = pd.read_csv("orders.csv", parse_dates=["date"])
df["quarter"] = df["date"].dt.to_period("Q").astype(str)

summary = (df.groupby(["region", "quarter"])
             .agg(total = ("amount",   "sum"),
                  count = ("order_id", "count"),
                  avg   = ("amount",   "mean"))
             .reset_index())

print(summary.head(10))

summary.to_excel("regional_summary.xlsx", index=False)