We used the sklearn Python package’s implementation of random forests, RandomForestClassifier. Our strategy of withholding some data in each training round is commonly referred to as leave-one ...