Marginal integration M-estimators of additive models

BOENTE, GRACIELA; MARTÍNEZ, ALEJANDRA

Marrakesh

Congreso; 61st ISI World Statistics Congress; 2017

International Statistical Institute

Partial linear additive models generalized the linear models since they modelate the relation between a response variable and covariates by assuming that some covariates are supposed to have a linear relation with the response but each of the others enter with an unknown univariate smooth functions. In this paper we propose two family of robust estimators for partial linear additive models. The first family is based on a three-step procedure which uses marginal integration for estimating the additive components and robust linear regression estimates for estimating the regression coefficient. The second approach of robust estimators combines B−splines with robust linear regression estimators. In order to compare the robust proposals between them and with their classical counterparts, a simulation study has been carried out under different partly linear additive models and different contamination schemes.