As a primarily full-stack TS dev, I've seen first-hand how difficult it is to perform simple ML-related tasks that the Python community takes for granted.
So I decided to try an experiment: how well would it work to auto-generate a TypeScript wrapper around Python's scikit-learn package, using a Python subprocess and IPC to communicate back and forth between them?
Turns out it works really well in practice, and so scikit-learn-ts was born.
The library supports all 257 classes exported by the official Python scikit-learn, and their TS wrappers are all auto-generated from the Python API docs.
This project's a lil hacky, but I've already reached out to the scikit-learn team, and we may experiment with using Pyodide as an alternative backend for the Python bridge.
Looking for feedback & ideas to improve the project. Thanks! :)