Robust estimation of multiple linear regression parameters in the presence of a problem of heterogeneity of variance and outliers values

Abstract

Often times, especially in practical applications, it is difficult to obtain data that is not tainted by a problem that may be related to the inconsistency of the variance of error or any other problem that impedes the use of the usual methods represented by the method of the ordinary least squares (OLS), To find the capabilities of the features of the multiple linear models, This is why many statisticians resort to the use of estimates by immune methods Especially with the presence of outliers, as well as the problem of error Variance instability, Two methods of horsepower were adopted, they are the robust weighted least square(RWLS)& the two-step robust weighted least square method(TSRWLS), and their performance was verified by applying it to simulate & selection the best methods for estimation by using measures mean absolute percentage error (MAPE) to compare them, the results show the method of (TSRWLS) is the best