ehrapy
ehrapy copied to clipboard
Support FAMD
Description of feature
This is very important because it's very common that categorical + numerical values are mixed. Also noticed that in Lennart's project. We should at least provide FAMD and provide guidelines on when to use which embedding methods.
https://github.com/MaxHalford/prince
- [ ] Get prince FAMD to run on MIMIC-II and store the results
- [ ] Implement FAMD with numpy for dense matrices and verify results
- [ ] Implement FAMD in faster with numba
- [ ] Implement FAMD with single dispatch also for sparse (https://github.com/theislab/ehrapy/pull/835)
- [ ] Implement FAMD with single dispatch also for dask
- [ ] Test everything in-depth