What is permutation importance in ensemble learning?

Permutation importance measures feature relevance by evaluating performance drop after shuffling feature values.

Permutation importance is a model-agnostic interpretability technique. It works by randomly shuffling one feature at a time and measuring how much model performance decreases. A large drop indicates high importance. This method is more reliable than impurity-based metrics because it directly measures impact on predictive performance. However, it can be computationally expensive and unstable with correlated features.