Método alternativo para regressão trivariada em presença de outliers

Abstract

This article discusses the estimation of the coefficients of the regression equation trivariada by a non-parametric method and its comparison with the classical method of least squares. Trivariados was simulated models with the presence of violations of the basic assumptions of regression analysis by the method of least squares. For performance evaluation of the mean square error methods was used. It was concluded that in this scenario the non-parametric method suggested has lower mean square error.