GradientBoostingRegressor

#include <Skigen/Ensemble>

template <typename Scalar = double>
class Skigen::GradientBoostingRegressor(loss=Loss::SquaredError, learning_rate=0.1, n_estimators=100, subsample=1.0, criterion=CriterionGB::FriedmanMSE, min_samples_split=2, min_samples_leaf=1, min_weight_fraction_leaf=0, max_depth=3, min_impurity_decrease=0, random_state=std::nullopt, alpha=0.9, verbose=0, max_leaf_nodes=std::nullopt, warm_start=false, validation_fraction=0.1, n_iter_no_change=std::nullopt, tol=1e-4, ccp_alpha=0)

Gradient Boosting for regression.

Stage-wise additive model that fits a sequence of regression trees on the pseudo-residuals of the current ensemble prediction. The final predictor is

$$\hat{F}_M(x) = F_0 + \eta \sum_{m=1}^{M} h_m(x)$$

where $F_0 = \bar{y}$ initialises the ensemble at the marginal mean (sklearn's init="zero" default uses a DummyRegressor that fits the mean — Skigen mirrors that behaviour exactly), $\eta$ is the learning rate, and each $h_m$ is a DecisionTreeRegressor trained on the residuals of stage $m-1$.

Mirrors sklearn.ensemble.GradientBoostingRegressor.

---

---

Attributes:

• loss : Loss

• learning_rate : Scalar

• n_estimators : int

• max_depth : int

• subsample : Scalar

• init : Scalar

• n_estimators_fitted : int

• estimators : const std::vector< DecisionTreeRegressor< Scalar > > &

• feature_importances : RowVectorType

• train_score : VectorType

---

Methods

SKIGEN_PARAMS()

---

predict(X)

---

score(X, y)

---