Generalized linear mixed models based on boosting

Abstract

A likelihood-based boosting approach for fitting generalized linear mixed models is presented. In contrast to common procedures it can be used in highdimensional settings where a large number of potentially influential explanatory variables is available. Constructed as a componentwise boosting method it is able to perform variable selection with the complexity of the resulting estimator being determined by information criteria. The method is investigated in simulation studies and illustrated by using a real data set. © 2010 Springer-Verlag Berlin Heidelberg.