Ranking and Selection Procedures for Bernoulli and Multinomial Data

Statistical analysis;Simulation;Output analysis

Malone, Gwendolyn Joy ; Industrial and Systems Engineering

University:Georgia Institute of Technology

Department:Industrial and Systems Engineering

关键词: Statistical analysis;    Simulation;    Output analysis;   

Others  :  https://smartech.gatech.edu/bitstream/1853/7603/1/malone_gwendolyn_j_200412_phd.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Ranking and Selection procedures have been designed to select the best system from anumber of alternatives, where the best system is defined by the given problem. The primaryfocus of this thesis is on experiments where the data are from simulated systems. In simulationranking and selection procedures, four classes of comparison problems are typicallyencountered. We focus on two of them: Bernoulli and multinomial selection. Therefore, wewish to select the best system from a number of simulated alternatives where the best systemis defined as either the one with the largest probability of success (Bernoulli selection)or the one with the greatest probability of being the best performer (multinomial selection).We focus on procedures that are sequential and use an indifference-zone formulationwherein the user specifies the smallest practical difference he wishes to detect between thebest system and other contenders.We apply fully sequential procedures due to Kim and Nelson (2004) to Bernoulli datafor terminating simulations, employing common random numbers. We find that significantsavings in total observations can be realized for two to five systems when we wish to detectsmall differences between competing systems. We also study the multinomial selectionproblem. We offer a Monte Carlo simulation of the Bechhofer and Kulkarni (1984) MBKmultinomial procedure and provide extended tables of results. In addition, we introduce amulti-factor extension of the MBK procedure. This procedure allows for multiple independentfactors of interest to be tested simultaneously from one data source (e.g., one personwill answer multiple independent surveys) with significant savings in total observationscompared to the factors being tested in independent experiments (each survey is run withseparate focus groups and results are combined after the experiment). Another multi-factormultinomial procedure is also introduced, which is an extension to the MBG procedure due to Bechhofer and Goldsman (1985, 1986). This procedure performs better that any otherprocedure to date for the multi-factor multinomial selection problem and should always beused whenever table values for the truncation point are available.