Getting Started with Polars
summary
Workshop uses Polars 0.17.5 on a real Twitter CSV (~5700 rows × 40 cols) and shows concrete moves: parse messy datetimes with strptime and timezone convert, cast large ints to u8/u16/u32 to cut memory, build features (is_reply, length, hour), and pivot/groupby aggregations. Benchmarks show a typical CSV read+transform at ~2.7 ms in Polars versus ~38 ms in pandas and expressions vs apply yields about a 10× speedup. It provides a runnable notebook and labs and demonstrates lazy evaluation and query planning.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:05Matts Background
- 8:23Loading Data
- 15:18Exploring Data Frame
- 17:17Exploring Data Types
- 20:49Lowering Memory Requirements
- 23:18Expressions
- 26:40Columns
- 29:03Vertical Stacking
- 31:57Transposing
- 35:02Integers
- 37:50Polar Share
- 46:17Polar Documentation
- 48:16Avoid Drop
- 49:38Renaming columns
- 51:14PLAll
- 52:04Task
- 55:33Troubleshooting
related talks
01h 32mAn Introduction to Polars
33mGet your Data prod ready, Fast, with Ruby Polars!
22mDon’t Write Polars Code with a Pandas Accent
37mPolars 1.0 and beyond
30mand time zones: everything you need to know
36mUnderstanding Polars Expressions when you're used to pandas
26mWe rewrote tsfresh in Polars and why you should too
26mLife Beyond Pandas: Workflows with DuckDB, Daft, Polars, and Datafusion