Doubly-nonparametric generalized additive models

Abstract

The popular generalized additive model framework is extended to allow both the mean curves and the response distribution to be nonparametric. The approach is demonstrated to be a flexible yet parsimonious tool for data analysis in its own right, as well as being a useful tool for model selection and diagnosis in the classical generalized additive model framework. Finite-sample performance of the method is examined via various simulation settings and the method is illustrated on two data analysis examples.