Equal size sampling

Hi folks, I’m doing is a decision tree but I don’t know if is necessary to insert a node of equal size sampling before the decision tree learner, decision tree predictor or both?


1 Like

HI @drmayorga -

What kind of problem are you working on? If you are doing a simple 0/1 classification, one thing you can do is to use stratified sampling on your target variable within the Partitioning node. This will ensure that you maintain an approximately proportional distribution for your target across your test and training datasets.

Does that help?