topic Re: extract rules from random forest model in SAS Data Science

extract rules from random forest model

stellarr — Fri, 11 Jun 2021 16:39:19 GMT

I have a trained model with hpforest proceduce and saved it as a binary file.
Is it possible to extract the rules of the model ?

Re: extract rules from random forest model

ballardw — Fri, 11 Jun 2021 20:14:37 GMT

It might help someone that knows more about this topic to show the code that created your model and how you saved the model.

Sometimes different options limit what may be stored. So if you have used such that person that understands the topic better than I may suggest changes to the code to rebuild the model in a way that will allow you to do what you request.

Re: extract rules from random forest model

stellarr — Fri, 11 Jun 2021 21:01:55 GMT

Thanks for the suggesting, below is the code. I'm using SASPY which uses thePROC HPFOREST

The saved model is the rf.bin binary file. The goal is to extract the rules from the rf.bin.Just like in decision tree we can WHOLETREEPLOT for HPSPLIT .

hpforest_model = ml.hpforest(data=hmeq, 
                             input=var_inputs, 
                             target="BAD /level = binary",
                             score = "out = work.forest_score",
                             save = "rf.bin",
                             id = "_PartInd_",
                             partition="rolevar=_partind_(TRAIN='1' VALIDATE='0');")