objectif

Boosting d’arbres pour signaux non linéaires.

code minimal

from sklearn.ensemble import GradientBoostingClassifier
from sklearn.datasets import load_iris
X, y = load_iris(return_X_y=True)
gb = GradientBoostingClassifier(random_state=0).fit(X, y)
print(hasattr(gb, "predict_proba"))

utilisation

from sklearn.ensemble import HistGradientBoostingRegressor
print(hasattr(HistGradientBoostingRegressor(), "fit"))

variante(s) utile(s)

from sklearn.inspection import permutation_importance
r = permutation_importance(gb, X, y, n_repeats=3, random_state=0)
print(r.importances_mean.shape[0] == X.shape[1])

notes

• HistGradientBoosting est plus rapide sur grands datasets.