Robust estimators for additive models with missing responses

BOENTE, GRACIELA; MARTÍNEZ, ALEJANDRA; SALIBIAN-BARRERA, MATÍAS

Oviedo

Conferencia; Fifth International Conference of the ERCIM Working Group on Computing & Statistics (ERCIM 2012); 2012

European Research Consortium for Informatics and Mathematics, International Association for Statistical Computing, Universidad de Oviedo, Queen Mary University of London

Additive models are widely used to avoid the difficulty of estimating regression functions of several covariates without using a parametric model(this problem is known as the curse of dimensionality). Different estimation procedures for these methods have been proposed in the literature, and some of them have also been extended to the case of data sets with missing responses. It is easy to see that most of these estimators can be unduly affected by a small proportion of atypical observations, and thus we are interested in obtaining robust alternatives. We consider robust estimators for additive models with missing responses based on local kernel M-estimators, and we also study a robust approach using marginal integration. If time permits, we will introduce a robust kernel estimator for additive models via the back-fitting algorithm.