A Generalized Goal in Restricted Subset Selection Theory
Abstract
A class of multiple decision procedures which selects a random size subset of populations not exceeding m (determined a priori) in size has previously been considered for the problem of selecting the best population among k candidates. The present paper generalizes the earlier results to the goal of selecting at least one of the t best populations. Applications of the basic theory can be made to many specific problems considered in the literature. In the regular case these include selection from univariate normal populations for means and variances, from gamma populations for scale factors and from noncentral chi square or noncentral F populations for noncentrality parameters and in the nonregular case selection from uniform populations with different supports.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1974
- Accession Number
- AD0780892
Entities
People
- Thomas J. Santner
Organizations
- Cornell University