Estimating Additive Models with Missing Responses

BOENTE, GRACIELA; MARTINEZ, ALEJANDRA

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS

TAYLOR & FRANCIS INC

Lugar: Londres; Año: 2016 vol. 45 p. 413 - 429

0361-0926

For multivariate regressors, the Nadaraya?Watson regression estimator suffers from the well?known curse of dimensionality. To overcome this drawback, additive regressionmodels have been introduced. All the procedures developed, up to now, to estimate the components under an additive model, assume that we observe all the data. However, in many applied statistical analysis missing data occur. In this paper, we study the effect of missing responses on the estimation of the regression function, under an additive regression model. The estimators are based on marginal integration adapted to the missing situation. The proposed estimators turn out to be consistent under mild assumptions. A simulation study allows to compare the behaviour of the our procedures, under different scenarios.