Some Contributions to Sequential Selection and Ranking Procedures,
Abstract
In Chapter 1, a class of sequential selection procedures is constructed for the selection of a unique binomial population as the best. The procedures hold good for the case when d is not specified. A class of sequential selection rules are also derived for the selection of the largest normal mean. A sufficient condition is given for the construction of rule which selects a unique population under subset selection formulation. P*-condition is satisfied for all cases. Confidence interval for the largest parameter and the bounds of asymptotic expected sample sizes are also studied. In Chapter 2, a problem of partitioning k normal populations with respect to a control is studied. In Chapter 3, the k-armed-bandit problem of finite sampling is formulated as a game. A maximin strategy is investigated. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1972
- Accession Number
- AD0747240
Entities
People
- Wen-tao Huang
Organizations
- Purdue University