cmaRs

An implementation of 'Conic Multivariate Adaptive Regression Splines (CMARS)' in R. See Weber et al. (2011) CMARS: a new contribution to nonparametric regression with multivariate adaptive regression splines supported by continuous optimization, doi:10.1080/17415977.2011.624770. It constructs models by using the terms obtained from the forward step of MARS and then estimates parameters by using 'Tikhonov' regularization and conic quadratic optimization. It is possible to construct models for prediction and binary classification. It provides performance measures for the model developed. The package needs the optimisation software 'MOSEK' to construct the models. Please follow the instructions in 'Rmosek' for the installation.