ABSTRACT
This chapter explains the additive model for multiple regression data and the backfitting algorithm for its estimation. It discusses commonly used additive model, the additive logistic model for binomial response data. The local-scoring algorithm can be applied to the additive analogue of generalized linear models, the class of generalized additive models. The interpretation problem highlights an important feature of the linear model that has made it so popular for statistical inference: the linear model is additive in the predictor effects. The linear model is used for regression in a variety of contexts other than ordinary regression. Common examples include log-linear models, logistic regression, the proportional-hazards model for survival data, models for ordinal categorical responses, and transformation models. Generalized Linear Interactive Modelling (GLIM) is a computer package for fitting generalized linear models. Iterative reweighted regression is the algorithm used in the GLIM package, an interactive computer program for fitting generalized linear models.
