Subset Selection Procedures Based on Some Robust Estimators

  • Song, Moon-Sub (Department of Computer Science & Statistics, Seoul National University) ;
  • Chung, Han-Yeong (Department of Computer Science & Statistics, Seoul National University) ;
  • Bae, Wha-Soo (Department of Computer Science & Statistics, Seoul National University)
  • Published : 1982.12.01

Abstract

In this paper, a preliminary study is performed on the subset selection procedures which are based on the trimmed means and the Hodges-Lehmann estimator derived from the Wilcoxon test. The proposed procedures are compared to the Gupta's rule through a small smaple Monte Carlo study. The results show that the procedures based on the robust estimators are successful in terms of efficiency and robustness.

Keywords