Comparison of Parameter Estimation Methods in Multiple Linear Regression Model When Data Contain Outliers in Independent Variables
Article Sidebar
Main Article Content
Abstract
The purpose of this study was to compare the methods of regression coefficient estimation in the multiple linear regression model for three methods, the ordinary least squares (OLS), M-Andrews and GM-Huber methods, in case of outliers on independent variable and error is t-distribution. The approach of estimator comparison was mean square error (MSE). The dataset was generated by the Monte Carlo simulation technique, repeated 1,000 times for each situation with R programming. The study results were as follows: the ordinary least squares methods indicated the lowest MSE when the degree of freedom equal to 5 in case of no outliers on independent. While the case of outliers on the independent variable, the GM-Andrews methods indicated the lowest MSE when the degree of freedom equal to 3 for all sample sizes and all percentage of outliers. However, the degree of freedom equal to 5, the M-Andrews methods indicated the lowest MSE in most cases.
Article Details
References
Huber, P.J., 1964, Robust estimation of a location parameter, Ann. Math. Stat. 35: 73-101.
Simpsona, J.R. and Montgomery, D.C., 1996, A biased-robust regression technique for the combined outlier multicollinearity problem, J. Stat. Comput. Simul. 56: 1-22.
Tantrakul, O. and Payakkapong, P., 2012, Comparison of robust regression methods in multiple linear regression model, Nat. Grad. Res. Conf. 13(1): 291-300. (in Thai)
Thitacharee, K., Sinsomboonthong, J. and Supapakorn, T., 2018, Robust comparison of regression coefficient estimation methods for multiple linear regression model when dataset contains outliers, Univ. Thai Chamber Commerce 2(1): 1686-1699. (in Thai)
Montgomery, D.C., Peck, E.A. and Vining, G.G., 2012, Introduction to Linear Analysis, 5th Ed., John Wiley & Son, Inc., New York.
Holland, P.W. and Welsch, R.E., 1977, Robust regression using iteratively reweight least-squares, Commun. Stat. Theory Meth. 6: 813-827.