The GAMSELECT Procedure

The GAMSELECT procedure fits and performs model selection for generalized additive models in SAS Viya.

Generalized additive models are extensions of generalized linear models. They relax the generalized linear models’ assumption of linearity by allowing spline terms that characterize nonlinear dependency structures. PROC GAMSELECT supports model selection by componentwise functional gradient descent (Friedman 2001; Bühlmann and Hothorn 2007) and the shrinkage method by using the penalized likelihood with sparsity-inducing penalties (Meier, Van de Geer, and Bühlmann 2009). It fits models for standard distributions in the exponential family, such as the normal, Poisson, gamma, and binary distributions.