SiDOR Group. University of Vigo.
devtools::install_github('sidoruvigo/wsbackfit')Documentation for package ‘wsbackfit-package’ version 1.0
Nonparametric statistical methods for estimating generalized additive and partially linear models as well as generalized varying coefficient models. Specifically, the package provides estimation procedures for a large class of regression models common in applied statistics. The regression models belong to the class of the so-called generalized structured models. They include for example generalized varying coefficient and generalized additive models, both optionally partial linear. The estimation procedure is based on smoothed backfitting which to our knowledge is the statistically most efficient existing procedure for this model class. Additional weights allow for both, the inclusion of different link functions and the efficient estimation under heteroscedasticity. Also, a cross validation bandwidth selector is provided.