I’ve been slowly building a collection of pure-NumPy (and a little SciPy) implementations of various ML models + building blocks to use for quick reference. The project has mostly been a fun thing for me to do in my spare time (hence the strange collection of models), though I hope it might also be useful for others interested in bare-bones implementations of particular models / ideas.

https://github.com/ddbourgin/numpy-ml

I’m sure there’s a ton that can be improved / made clearer. Alternatively, if you have models of your own that would be a good fit, PRs are welcome 🙂