Abstract

This entry provides an overview of multiple and generalized nonparametric regression from a smoothing spline perspective. Details are provided on smoothing parameter selection for Gaussian and non-Gaussian data, diagnostic and inferential tools for function estimates, function and penalty representations for models with multiple predictors, and the iteratively reweighted penalized least squares algorithm for the function estimation. Two different smoothing frameworks are compared: smoothing spline analysis of variance (SSANOVA) and generalized additive models (GAMs). Examples with supporting R code are provided.