Robust Estimators are Hard to Compute

Abstract

In modern statistics, the robust estimation of parameters of a re- gression hyperplane is a central problem. Robustness means that the estimation is not or only slightly a®ected by outliers in the data. In this paper, it is shown that the following robust estimators are hard to compute: LMS, LQS, LTS, LTA, MCD, MVE, Constrained M es- timator, Projection Depth (PD) and Stahel-Donoho. In addition, a data set is presented such that the ltsReg-procedure of R has proba- bility less than 0.0001 of ¯nding a correct answer. Furthermore, it is described, how to design new robust estimators.