Feature Importance Analysis

Mean Decrease Accuracy (MDA)

Importance evaluation technique measuring the performance drop when a feature is removed or permuted, using out-of-bag error in random forests.

← Indietro