0 votes
Hello all,

I use Algorithm  : Random forest classification in Dataiku.

But in confusion matrix, I still see the threshold cut-off, I think it is only for regression? Anyone can help me to explain ? and how to choose Random forest regression (output is probability)
asked by

1 Answer

0 votes

All Classification algorithms in DSS are meant to output classes (either binary: 0/1, or multi-class). The vast majority of classification algorithms don't directly predict classes but probabilities, and then apply a threshold on the probability.

Random Forest Classification is one of these, so it does predict a probability and then applies a threshold to it. If you deploy a Random Forest Classification model in DSS, it will output both the probability and the thresholded predicted class. Thus, if you are only interested in the probability, you don't need to bother about the threshold, and just use the predicted probabilities columns in the result.

In DSS, Random Forest Regression only applies to continuous scoring (ie predict a numerical variable like price, instead of a discrete variable like color).
answered by
742 questions
763 answers
484 users