// the find
jvns/pandas-cookbook
Recipes for using Python's pandas library
A set of Jupyter notebooks walking through pandas basics using three real-world datasets: NYC 311 calls, Montreal bike path counts, and Montreal weather. It's aimed at people who already know some Python but have never used pandas and find the official docs too dense to start with.
Uses real messy data instead of toy examples, so you actually hit encoding issues, inconsistent column types, and string cleanup the way you would in practice. The chapter split (selecting data, groupby/aggregate, string ops, timestamps, SQL loading) matches the order people actually get stuck in. Jupyter Lite support means you can run it in the browser with zero local setup, which matters for a cookbook aimed at beginners.
It's from ~2015-era pandas and hasn't been meaningfully updated since — no coverage of newer pandas APIs (e.g. the groupby/apply deprecations, nullable dtypes, pyarrow backend), and some patterns taught here are now discouraged. No tests or CI beyond a deploy workflow, so there's no guarantee the notebooks still execute cleanly against current pandas/numpy versions. It's a fixed set of nine chapters with no ongoing maintenance model, so it's a snapshot of 2015 pandas idioms rather than a living reference.