Local Estimators in Multivariate Generalized Lin- Ear Models with Varying-coeecients

The varying-coeecient model allows to investigate and visualize the form of the interaction of variables in the predictor. Moreover, common approaches like semiparametric modelling and generalized linear models are special cases. The focus is on local estimators, in particular local likelihood and locally weighted least squares estimators, which both are consistent and asymptotically normally distributed. Extensions to the case of multivari-ate smoothing are considered. An example illustrates the usefulness of the approach.