Robust Regression Methods / a Comparison Study

Main Article Content

Maha Issa Ismail; Huda Abdullah Rasheed

Abstract

The ordinary least squares method (OLS) is one of the most common methods for estimating
the coefficients of linear regression models. However, it is sensitive and not robust against
the existence of outliers. Therefore, several robust estimation methods have been used and
then represented by M-estimation using different objective functions. In this paper, a number
of alternative robust methods have been suggested that represented by using Gastwirth’s
location estimator instead of the mean in OLS and instead of the median in different Mestimation
methods. In addition to repeating the Hubers' M-estimation method (first method)
until converged results are reached. A Monte-Carlo simulation study was employed to
evaluate the performance of different estimation methods depending on the MSE of
regression coefficients.

Article Details

Section
Articles