23.7 Practice, Tip and Close

By Ravindra Bagale Last updated: 23 October 2026

Why. Mistakes compound at project time (Chapter 16). Fixing them here saves the Blinkit story.

What. One lab that plants bugs on purpose, then heals them.

How.

Chapter practice lab

Build a 12-row fictional orders frame with comma amounts, one N/A, mixed city case, and two stores sharing city Pune in a dimension table.
Demonstrate a chained assignment warning/fix; replace with .loc.
Clean dtypes; log NaNs from coerce.
Show a bad city-only merge explosion; fix with store_id + validate="m:1".
Localise two IST timestamps; compute minutes.
Filter Delivered in {Pune, Nashik} with parentheses; print before/after.
Run peek() before exporting CSV.

# Seed for the lab — expand and break it on purpose
import pandas as pd

orders = pd.DataFrame({
    "order_id": [f"BLK-{i}" for i in range(1, 7)],
    "store_id": ["PUN-01", "PUN-02", "PUN-01", "NAS-01", "NGP-01", "KOL-01"],
    "city": ["Pune", "pune", "Pune", "Nashik", "Nagpur", "Kolhapur"],
    "amount": ["100", "1,250", "90", "N/A", "80", "110"],
    "status": ["Delivered", "Delivered", "Cancelled", "Delivered", "Delivered", "Delivered"],
    "customer": ["Ravindra Bagale", "Zoya", "Amir", "Salman", "Raja", "Ravina"],
})
print(orders)

Practice task

After fixing the lab, export outputs/orders_fixed.csv and write three bullets: NaNs found, merge rows before/after, filter before/after.

Ravindra Bagale's Tip

pandas errors shikavtat. Warning ignore naka — to free teacher aahe.

Thodkyaat sangaycha tar (quick recap)

Use .loc — avoid chained assignment
Coerce junk to NaN and count it
Merge on true keys; watch row explosion
Be explicit about IST vs UTC vs date-only
Parentheses with & / |; log filter counts
Debug with shape, NA, dups, indicator, control totals

Samajla ka? Hya teen chapters ni stats, sharing ani pandas bugs tight kele — abhyas regular theva, Jupyter ughadun ek bug/day fix kara.