Global Interpretability

Permutation Feature Importance

Agnostic method that evaluates variable importance by measuring the degradation of model performance when a feature's values are randomly permuted, revealing their global contribution.

← Geri