Closed Adaptive Sequential Procedures for Selecting the Best of k > or = 2 Bernoulli Populations.
Abstract
The goal of selecting that one of k > or = 2 Bernoulli populations which has the largest single-trial 'success' orobability is treated. Consideration is restricted to procedures which take no more than n observations from any one of the K populations. One such procedure is the single-stage procedure of Sobel and Huyett (1957) which takes exactly n observations from each of the k population. We propose a one-at-a-time adaptive sampling rule (R*) which when used in conjunction with a particular stopping rule (S*) and terminal decision rule (T*) achieves the same probability of a correct selection as does the single-stage procedure uniformly. The procedure (R*, S*, T*) is generalized for k > 2 to accommodate such goals as 'Selecting the s (1 < or = s < or = k-1) 'best' Bernoulli populations with regard to order,' and is shown to have desirable properties for these goals as well.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1981
- Accession Number
- ADA115653
Entities
People
- Radhika V. Kulkarni
- Robert E. Bechhofer
Organizations
- Cornell University College of Engineering