Computerassisted Semiparametric Generalized Linear Models

The paper gives an overview on generalized linear models and its application in different branches of science. We introduce semiparametric extensions of the generalized linear model. One particular model of interest is the generalized partially linear model which allows a nonparametric modeling of the influence of the continuous covariables. The estimation procedure is introduced and a test on the correct specification of this model (vs. a parametric generalized linear model) is presented. The application to a data set on East-West German migration illustrates the use of this technique. Semiparametric methods are highly demanding on software. Interactive graphics to display the resulting curves/surfaces and tools for efficient computation on the user level are such requirements. We thus complete our presentation with the practical implementation in new version of the statistical computing environment XploRe which offers these features