G-Research, a quantitative finance research firm, shares testimonials from its machine learning engineers about switching from pandas to Polars for data processing. Users report significant speedups, including a 150x improvement in one case and a 20x speedup in dequantization, along with lower memory usage and a cleaner API. The firm highlights Polars' multi-threaded joins, responsive bug fixes from the development team, and improved developer experience compared to pandas, noting the library is now used across multiple internal projects including a plotting library.
Table of contents
About G-ResearchWhat Makes Polars So Special for G-Research?What Users at G-Research Are SayingThe Bottom Line: Polars is Changing the Game2 Impressions