objectif

Mesurer l’importance par permutation (agnostique au modèle).

code minimal

from sklearn.ensemble import RandomForestClassifier
from sklearn.inspection import permutation_importance
from sklearn.datasets import load_iris
X, y = load_iris(return_X_y=True)
rf = RandomForestClassifier(random_state=0).fit(X, y)
r = permutation_importance(rf, X, y, n_repeats=5, random_state=0)
print(len(r.importances_mean) == X.shape[1])

utilisation

print((r.importances_mean >= 0).all())

variante(s) utile(s)

from sklearn.linear_model import LogisticRegression
print(hasattr(LogisticRegression(), "fit"))

notes

• Plus coûteux; robuste aux biais des arbres.