Robust estimators under a semi-functional partly linear model

BOENTE, GRACIELA; VAHNOVAN, ALEJANDRA

Londres

Conferencia; VI International Conference of the ERCIM Working Group on Computing & Statistics (ERCIM2013).; 2013

Computational Statistics and Data Analysis

Partial linear modelling ideas have recently been adapted to situations in which functional data are observed. To be more precise, semi-functional partly linear models deal with the problem of predicting a real-valued response variable in the situation where some of the explanatory variables are functional. The aim of such a model is to allow the functional explanatory variables to act in a free nonparametric manner, while the multivariate ones are controlled by means of a parametric linear relation. Estimators for these models have been recently considered using a functional version of the Nadaraya-Watson estimators to deal with the nonparametric component. It is easy to see that these estimators can be affected by a small proportion of atypical observations, and hence the interest is to obtain robust estimators for semi-functional partly linear model. A robust approach is considered that combines robust local kernel M-estimators. with robust regression estimators.