Estimação dos parâmetros do modelo de regressão linear simples pelo método não-paramétrico

Abstract

The most used method to estimate the angular and linear coefficients of the line equation in simple linear regression model is the least square method, in spite do the existence of another. The nonparametric method, which is less used because its laborious process. In this context, the goal of this work is implement an algorithm, using the Visual Basic for Applications (VBA) which can became easier to use the nonparametric method to compare the estimates and criterias found by parametric method. The methodology used was the estimator of Theil’s estimator angular coefficient (b) and the two estimators proposed by Dietz for the intercept (a). At the end of research the algorithm implemented in VBA really makes easier the use of the nonparametric method in academic environment, because the estimates found by the two methods were very close, with small differences among the criterias used to compare the estimated equations.