Precision/recall from out-of-bag predictions in RandomForestClassifier

Question

I would like to use out-of-bag training/validation with a classifier such as RandomForestClassifier.

Is it possible to get the out-of-bag predictions?

I want the OOB predictions so I can compute precision/recall on instances not used to train the classifier.

If there are alternative approaches to getting the precision/recall, please let me know.

Answer 1

It is possible to use RandomForestClassifer.oob_decision_function_ to compute the out-of-bag estimate on the training set.