Ravindra BagaleCourses & study guides मराठी Track your progress

25. How to Rank and Running Totals in pandas (rank, cumsum, window)

25.4 Group-wise Rank and Cumsum

By , trainer LinkedIn Last updated:

Why. Global rank says Nashik’s best order is #40 statewide. Pune manager wants within Pune ranks. Same for running GMV per city.

What.

  • df.groupby("city")["amount_inr"].rank(ascending=False, method="dense")
  • df.groupby("city")["gmv"].cumsum() after sorting by city + date

How.

Within-city leaderboard + running GMV

  1. Sort by city, then order_date (or amount for ranks only).
  2. Add city_rank with groupby rank.
  3. Add city_running_gmv with groupby cumsum on a GMV column.
  4. Spot-check one city with .loc[df.city == "Pune"].
  5. Confirm each city’s running total resets (first Pune row ≠ continue from Nashik).

Worked example A — Rank inside each city.

import pandas as pd

df = pd.DataFrame({
    "order_id": ["BLK-1", "BLK-2", "BLK-3", "AMN-4", "BLK-5", "BLK-6"],
    "city": ["Pune", "Pune", "Pune", "Nashik", "Nashik", "Kolhapur"],
    "amount_inr": [210, 180, 210, 95, 120, 340],
    "customer": ["Ruhi", "Amir", "Zoya", "Ravina", "Salman", "Rani"],
})
df["city_rank"] = (
    df.groupby("city")["amount_inr"]
    .rank(ascending=False, method="dense")
)
print(df.sort_values(["city", "city_rank"]))

Worked example B — Running GMV per city.

daily = pd.DataFrame({
    "city": ["Pune", "Pune", "Pune", "Nashik", "Nashik"],
    "order_date": pd.to_datetime(
        ["2026-10-22", "2026-10-23", "2026-10-24", "2026-10-23", "2026-10-24"]
    ),
    "gmv": [50000, 62000, 71000, 22000, 25000],
})
daily = daily.sort_values(["city", "order_date"])
daily["city_running"] = daily.groupby("city")["gmv"].cumsum()
print(daily)

Worked example C — Top-1 per city.

top1 = df.loc[df["city_rank"] == 1, ["city", "order_id", "amount_inr"]]
print(top1)

Ravindra Bagale's Tip

Global rank vs city rank — donhi different answers. Manager konta pehle sanga.