mboost: Model-Based Boosting

• Authors: Torsten Hothorn and Peter Bühlmann with contributions by Thomas Kneib and Matthias Schmid
• mboost is an R-package implementing functional gradient descent algorithms (boosting) for optimizing general loss functions utilizing componentwise penalized least squares or regression trees as base learner for fitting generalized linear, additive and interation models to potentially high-dimensional data.
• Function gamboost implements model choice and variable selection for general geoadditive regression models comprising nonparametric effects of continuous covariates, interation surfaces and spatial effects, varying coefficient terms and random effects. Base-learners are constructed as penalized least squares fits with penalized splines being the main modelling component.
• Download: http://cran.r-project.org/