Unbiased variable importance for random forests
This addresses a specific issue for users of random forests who rely on variable importance measures, but it is incremental as it modifies an existing method rather than introducing a new paradigm.
The authors tackled the problem of biased variable importance measures in random forests by proposing a simple solution that computes loss reduction on out-of-bag samples instead of in-bag samples, resulting in an unbiased alternative to the default Gini importance.
The default variable-importance measure in random Forests, Gini importance, has been shown to suffer from the bias of the underlying Gini-gain splitting criterion. While the alternative permutation importance is generally accepted as a reliable measure of variable importance, it is also computationally demanding and suffers from other shortcomings. We propose a simple solution to the misleading/untrustworthy Gini importance which can be viewed as an overfitting problem: we compute the loss reduction on the out-of-bag instead of the in-bag training samples.