Hi All,
I'm trying to find some details on how the predictor screening platform operates, I know the controllable parameter is the number of trees, but what are the settings for the bootstrap forest that are being ran in the background?
I've found I can't easily recreate the output from Predictor Screening in the Bootstrap Forest for ranking of variables with the Bootstrap Forest default values. I know there will be some differences dependent on the number of iterations of each, but want to be able to closely match the predictor screening results.
Do we have any internal documents describing these differences?
Thanks,
Ben
