23. Common pandas Mistakes and Fixes

23.1 SettingWithCopy and Chained Assignment

By Ravindra Bagale · Last updated: 23 October 2026

Why. You filter Pune rows, then set delivery_mins = 12, and nothing sticks — or pandas warns. Shraddha loses twenty minutes thinking the dark store data “won’t update”.

What. Chained assignment means two __getitem__ steps in one line, like df[df.city == "Pune"]["mins"] = 12. pandas may give you a copy, so the write vanishes. Prefer .loc on the original frame, or work on an explicit .copy().

How.

Worked example A — Broken vs fixed.

import pandas as pd

df = pd.DataFrame({
    "order_id": ["BLK-1", "BLK-2", "BLK-3", "AMN-4"],
    "city": ["Pune", "Pune", "Nashik", "Kolhapur"],
    "delivery_mins": [18, 22, 11, 10],
    "customer": ["Ruhi Bagale", "Amir", "Zoya", "Ravina"],
})

# Fragile — may not write back
df[df["city"] == "Pune"]["delivery_mins"] = 12
print("after chain", df["delivery_mins"].tolist())

# Clear — writes on df
df.loc[df["city"] == "Pune", "delivery_mins"] = 12
print("after loc", df["delivery_mins"].tolist())

Worked example B — Explicit copy for a side analysis.

pune = df.loc[df["city"] == "Pune"].copy()
pune["sla_flag"] = pune["delivery_mins"] <= 15
print(pune[["order_id", "sla_flag"]])
# original df has no sla_flag — intentional
print("sla in df?", "sla_flag" in df.columns)

Worked example C — New column with np.where.

import numpy as np

df = df.copy()
df["late"] = np.where(df["delivery_mins"] > 15, "Late", "OK")
print(df[["city", "delivery_mins", "late"]])

Raja in Nagpur uses the same pattern for cancel flags — one column, no chain.