Explore the latest advancements in Dask DataFrame 2.0 and its integration with pandas in this 30-minute PyCon US talk. Discover how recent improvements address historical performance issues, making Dask a more robust and user-friendly option for big data processing. Learn about the new shuffle algorithm, logical query planning layer, and reduced memory footprint resulting from pandas 2.0. Compare Dask's capabilities to other popular big data tools like Spark, Polars, and DuckDB using TPC-H benchmarks. Gain insights into the future developments of pandas and Dask, including potential extensions of the logical query planning layer to frameworks such as Dask Array and XArray.
Overview
Syllabus
Talks - Patrick Hoefler: Pandas + Dask DataFrame 2.0 - Comparison to Spark, DuckDB and Polars
Taught by
PyCon US