How Do I Reconstruct A Random Forest Output?

Question

[Eminencegrise]

I am working on a classification problem. I have a set that has high AUC yet I am not sure how to uncover the cominatory rules that generated that specific output.

[Neilfws]

I think we'll need more details to answer this. In particular, it should be a bioinformatics problem, not a general machine-learning problem; otherwise http://stats.stackexchange.com is a better forum.

[Khader Shameer]

or Metaoptimize Q&A http://metaoptimize.com/qa/ ?

[Neilfws]

I think we'll need more details to anser this. In particular, it should be a bioinformatics problem, not a general machine-learning problem; otherwise http://stats.stackexchange.com/ is a better forum.

[Khader Shameer]

or Meta-optimize http://metaoptimize.com/qa/ ?

Answer 1

[David Quigley]

Khader's answer is a good pointer; I hadn't heard of that site, which focuses on machine learning, NLP, etc. See here, for example.

Answer 2

[Shreyas]

If you are using Random Forest in R try the varImpPlot function on your classifier object.