How to change the function a random forest uses to make decisions from individual trees?

Question

Random Forests use 'a multitude of decision trees at training time and outputting the class that is the mode of the classes (classification) of the individual trees'.

Is there a way to, instead of using the class that is the mode, run another random forest on the outputs produced by the original trees?

Bonus question: is there a reason why this is a bad idea? (as I'm sure people will have thought of this before)

Answer 1

You can access the individual decision trees in the estimators_ attribute of a fitted random forest instance.

You can even re-sample that attribute (it's just a Python list of decision tree objects) to add or remove trees and see the impact on the quality of the prediction of the resulting forest.

Answer 2

I assume is just a performance option, your idea sounds fine, but without better "randomness" but probably slower on being computed.