Robust estimators in Generalized Partially Linear Models

GRACIELA BOENTE, XUMING HE Y JIANHUI ZHOU

Madrid, España

Workshop; Workshop on Robustness and Statistical Inference, In Honour of Victor Yohai.; 2006

Universidad Carlos III de Madrid

In this talk, we will first remind the classical approach to generalized partly linear models. The sensitivity to outliers of the classical estimates for these models is good evidence that robust methods are needed. The problem of obtaining a family of robust estimates is then addressed introducing a general two–step robust procedure to estimate the parameter and the regression function, under a generalized partly linear model (1). The asymptotic behavior of the robust estimates of and the regression function, under a generalized partly linear model (1). The asymptotic behavior of the robust estimates of is derived. Through a Monte Carlo study, the performance of the proposed estimates is compared with the classical ones. A robust procedure to choose the smoothing parameter is also discussed.