Trying polars, a fast DataFrame library
pandas is the de facto standard DataFrame manipulation tool in Python.
Natually, I want to be comfortable using it but, I think the sheer girth of its API is proving to be a major obstacle. I can't go for long before I have to search usages on Stackoverflow for what should be routine tasks. I don't think I can ever grok its date time API.
So, I searched for simpler alternative libraries and found polars. It is written in Rust and uses the Arrow format for memory model. Its performance benchmarks are impressive, so apparently the newfangled technology stack is pulling its weight.
The API seems still in development and the features are miniscule compared to pandas's, but I appreciate its simplicity. Following is a notebook of me trying to use polars for one of the tutorials on Kaggle Courses.