Comment by ralegh

I personally found the polars API much clunkier, especially for rapid prototyping. I use it only for cemented processes where I could do with speed up/memory reduction.

Is there anything specific you prefer moving from the pandas API to polars?