weak learner

[GoogleCodeExporter]

hi 

I just find this implementation of randomforest in matlab,
I want to know if it is possible to modify the algorithm ?
what I want is to change the weak predictor ? is it possible to have specific 
split function for each node that is defined by user or one should use the tool 
as a black boxed function call ?

thanks

Original issue reported on code.google.com by [email protected] on 18 Dec 2012 at 8:45

[GoogleCodeExporter]

the algorithm is somewhat complicated so it may non-trivial to change the split 
function. The split function is called findBestSplit and is in a .f file for 
classification and in regRf.cpp for classification.

yup, rf is more or less used like a black box function call.

Original comment by abhirana on 19 Dec 2012 at 9:10

• Changed state: Accepted