On Robust Alternatives to the Maximum Likelihood Estimators of a Linear Functional Relationship

This paper discusses a number of candidates for robust estimator TS of a simple linear functional relationship (SLFR). The classical maximum likelihood estimators of the SLFR can be affected by the presence of possible outliers. This is due to the fact that they are mean-based estimators. Some median-based estimator of the SLFR are examined. Among those considered Their-type estimators and the modified LI-norm estimator are found to be most robust (i.e insensitive to the outliers).

Preview PDF On_Robust_Alternatives_to_the_Maximum_Likelihood_Estimators.pdf Download (4MB)

Actions (login required)

View Item