Abstract

Wenceslao Gonzalez Manteiga, University de Santiago de Compostela

"Smoothing techniques for regression with directional data"

The analysis of directional data has gained attention in recent years, with the proposal of different statistical methods for the study of density and regression curves involving directional variables. In this talk, we will consider a location-scale regression model with linear response and directional predictor. From a nonparametric perspective, estimation and testing problems on this model will be approached. First, a local-linear estimator for the regression curve will be proposed, and later, this estimator will be used as a pilot for assessing the goodness-of-fit of a parametric model. Performance of the testing procedure, in terms of size and power, will be explored and illustrated with simulated and real data. This is a joint work with E. García-Portugués, I. Van Keilegom and R.M. Crujeiras.