SELECCION DE VARIABLES EN LA REGRESION LINEAL CON EL ALGORITMO RRQR RESTRINGIDO

Abstract

The polemical problem of variable selection has originated different procedures when seeking for the regression equation that best fits the data with minimun number of parameters. In this paper a new procedure for variable selection is proposed, which combines the restricted RRQR algorithm with the statistical criterium of Mallows for model selection. Two applications illustrate the advantages of this procedure.