pyMCR
pyMCR copied to clipboard
Rank submodule
Submodule to estimate rank of the data.
Methods to incorporate (so far):
- SVD/PCA + various metrics (I do a lot of these)
- HySime
- SIMPLISMA (@ClarkAH)
- IND and RSD (@sshojiro )
Implementation considerations
- Out-of-core analysis
- Python vs Cython for some
- ?
Note that rank approximation is not a trivial task. It's super-easy to get an estimate and seemingly easier to be incorrect. Kind of the reason I didn't incorporate this feature into the 0.3.0 release.
My plan is to create a super class RankEst that all subclasses will inherit. This class will likely incorporate some metrics to describe confidence or error -- not sure.
Questions and comment welcome!