Implement MORF

[eigenvivek]

Is your feature request related to a problem? Please describe.

See issues #23, #64, and #65. A base implementation of MORF would be useful and is necessary for additional MORF studies.

Describe the solution you'd like

Implement ManifoldObliqueSplitter and ManifoldObliqueTreeClassifier in transformers.py.

Should be very simple to make the splitter. The class can inherit everything from ObliqueSplitter, except for the sample_proj_mat function. That needs to be updated to use the sampling function for MORF.

Describe alternatives you've considered

Additional context (e.g. screenshots)

[eigenvivek]

hi @mkusman1, having a working morf implementation is mission critical for experiments people in the lab are planning to run

im going to prioritize getting this done in the next two weeks (see #418)

#65 is a downstream issue from this that @bstraus1 was interested in. im sure there are others.

one thing that would be useful is having a way of measuring feature importance in a forest. this is implemented in the original SPORF repo in R (see here), and it would be useful to have this in proglearn. @PSSF23 can you confirm that this is not already implemented in proglearn?

implementing feature importance would be a fine thing to do for the spring, probably along with recreating this tutorial notebook from the original repo.

[mkusman1]

@PSSF23 @v715 Okay I perfectly understand that!

In this case, I would be happy to work on implementing feature importance. Should I create a new issue for that?

[PSSF23]

@v715 I can confirm that feature importance is absent from the current implementation.

@mkusman1 Always feel free to open new issues after confirming no repetition~