bst-package


Gradient descent boosting for hinge loss and square error loss.


Functional gradient descent algorithm for a variety of convex and nonconvex loss functions, for both classical and robust regression and classification problems. HingeBoost is implemented for binary and multi-class classification, with unequal misclassification costs for binary case. The algorithm can fit linear and nonlinear classifiers.

bst-package: Boosting for Classification and Regression

Description

Arguments

Details