Standard Article

Subset Selection

Statistical and Numerical Computing

  1. Paul Van Der Laan

Published Online: 15 SEP 2006

DOI: 10.1002/9780470057339.vas063

Encyclopedia of Environmetrics

Encyclopedia of Environmetrics

How to Cite

Van Der Laan, P. 2006. Subset Selection. Encyclopedia of Environmetrics. 6.

Author Information

  1. Eindhoven University of Technology, The Netherlands

Publication History

  1. Published Online: 15 SEP 2006


Assume k(k ≥ 2) populations are given. The associated independent random variables have continuous distribution functions differing only in their unknown location parameter. The statistical selection goal of subset selection is to select from the given set of k populations a non-empty subset of these populations so that the selected subset includes with confidence level P*, with k−1 < P* < 1, the best population. In this context, best population means the population with the largest value of the location parameter. The subset selection approach is described in some detail and some properties are presented.


  • selection;
  • probability of correct selection;
  • expected subset size;
  • variance of subset size;
  • least favourable configuration