Accession Number : AD0747236

Title :   On a Maximum Strategy for Sampling Based on Selection Procedures from Several Populations,

Corporate Author : PURDUE UNIV LAFAYETTE IND DEPT OF STATISTICS

Personal Author(s) : Gupta,Shanti S. ; Huang,Wen-Tao

Report Date : JUL 1972

Pagination or Media Count : 27

Abstract : Let Pi 1, Pi 2,..., Pi(k) be k populations such that Pi(i) has cdf F(x;theta sub i). It is permitted to draw n samples from these k populations and the sum of these n numerical samples is considered to be the reward. It is desirable to make the expected reward as large as possible. The problem is formulated as a game based on a selection procedure. The maximin strategy of the game formulated is used for the sampling scheme. Some properties of selection rules advantageous for the sampling scheme are studied. Sufficient conditions for the existence of the value of the game are given. The least favorable configuration for the game is found when the populations are normal with common known variance. An asymptotic optimals property for the maximin strategy is shown to hold. (Author)

Descriptors :   (*SAMPLING, OPTIMIZATION), (*GAME THEORY, OPTIMIZATION), PROBABILITY, RANDOM VARIABLES, CONVEX SETS, THEOREMS, TABLES(DATA)

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE