A Comparative Study of Robust Regression Methods in Modeling the Currency in Circulation in Malaysia

This study presents a comparative study on robust regression method via M-estimator and Least Trimmed Square (LTS) estimator against Ordinary Least Square (OLS) estimator in modelling economic indicators in Malaysia. The aim is to estimate the parameters of multiple linear regression model in the presence of outliers and to evaluate the estimator performance of OLS, LTS and M estimation. The coefficient of determination known as R-squared values were used as criteria to evaluate the estimator performance. Research findings showed that LTS estimator outperform M estimator and OLS estimator in terms of maximum of R-squared values.