DuckDB is the New Pandas
summary
Claims that DuckDB replaces Pandas with concrete evidence: a 17M-row groupby distinct on parcels runs 0.2s vs 4s (≈20–25× faster), uses all cores, and reads remote files via HTTPFS and Parquet on S3 to query a 1 TB / 16M-judgments dataset without downloading. Complex analytics (bench-wise average disposal time) runs against remote parquet, LLMs generate SQL scaffolding, and DuckDB embeds in Python/Node/wasm for interactive grading in the browser — Wasm browser embed demonstrates it actually works.
AI-written summary · about
chapters (12)
- 0:00Introduction to DuckDB
- 1:30Why switch from Pandas
- 4:49Performance comparison with Pandas
- 6:16Querying remote files
- 7:48Processing large datasets
- 9:10Using LLMs for SQL queries
- 12:50Analyzing court judgments
- 18:43Anagram detection with DuckDB
- 22:45Embedding DuckDB everywhere
- 24:55Running DuckDB in browser
- 26:03Testing DuckDB with ChatGPT
- 27:06Extensions and future use
