Providing for the Analysis of Generalized Additive Models within a System already Capable of Generalized Linear and Nonlinear Regression

Additive models are a relatively new addition to the practical statistician’s toolkit. They provide an appealing alternative to parametric models, allowing relationships between variables to be explored without imposing pre-defined constraints on the form or shape of the relationships. They can be used in the initial stages of analysis to let the data suggest parametric forms, and they can be used throughout to eliminate effects that do not need to be parameterized. Additive effects can be included in a model together with parametric forms where relevant, and incorporated in generalized linear models or nonlinear models as well as in linear models.