How to make spark decisiontree model use feature subsetting?

28 Views Asked by At

I am trying to build a random forest model using pyspark ml library. However, there is some special bootstrapping strategy that fits my dataset. So my plan is to do the bootstrapping separately and then train a bunch of decision tree models, bagging them on my own as a random forest model. Here comes the problem: it seems that feature subsetting in spark decision tree is only reserved for random forest (this is my understanding of the source code). Is there any way to enable such behavior instead of some work-around like training multiple random forest with numTrees=1?

0

There are 0 best solutions below