search

manifoldai / merf

Mixed Effects Random Forest
MIT License
219 stars 52 forks source link

Add partial dependency plotting function #26

Closed OliverFishCode closed 4 years ago

OliverFishCode commented 5 years ago

It would be nice if there was a function for creating partial dependency plot data or plots. This would help with translating information to consumers and clients

resdntalien commented 5 years ago

This sounds like a good idea, but I'm not exactly sure what you mean. A great thing to contribute if you have the time!

OliverFishCode commented 5 years ago

This is how is commonly implimented in R : https://www.rdocumentation.org/packages/randomForest/versions/4.6-14/topics/partialPlot

This is the implementation in sci-kit learn: https://scikit-learn.org/stable/auto_examples/ensemble/plot_partial_dependence.html

I'm not sure how it would work incorporating random effects, but I'm sure it would be similar to BLUP (best linear unbiased predictor) or no BLUP in GLMM (generalized linear mixed models). I may take a look at this sometime ( my plates a little full), but I am not sure if the math and code is beyond my ability ( I have a solid case of imposter syndrome).

resdntalien commented 4 years ago

@OliverFishCode I am closing this for the time being. Please see the MERF example notebook of how to do this by accessing the internal trained RF. This does not however address doing this WITH the random effect applied as well. If you want to do that I think you would have to do it per cluster.... Unclear.