A Bayesian batch means methodology for analysis of simulation output

Authors:
Richard W. Andrews;Thomas J. Schriber
Affiliations:
Graduate School of Business, The University of Michigan, Ann Arbor, MI;Graduate School of Business, The University of Michigan, Ann Arbor, MI
Venue:
WSC '83 Proceedings of the 15th conference on Winter simulation - Volume 1
Year:
1983

Citing 1
Cited 2

A conceptual framework for research in the analysis of simulation output

Communications of the ACM - Special issue on simulation modeling and statistical computing

Bayesian analysis for simulation input and output

Proceedings of the 29th conference on Winter simulation
Selecting the best system: a decision-theoretic approach

Proceedings of the 29th conference on Winter simulation

Quantified Score

Hi-index	0.00

Visualization

Abstract

The purpose of this research is to investigate the use of Bayesian methodology in the analysis of simulation output. Specifically, the Bayesian methodology is introduced in the context of the batch means procedure for building a confidence interval for the output mean. We assume that the output process is at steady state or equivalently that the output process is second order stationary. We also assume that the length is fixed at say n. So the output process can be given by X1, X2, ..., Xn,a sequence of observations from a continuous state stationary stochastic process with mean M, variance O2x and autocorrelation function {Pi}@@@@I&equil;l This Bayesian batch means methodology has been thoroughly tested. The five measures of effectiveness suggested in Schriber and Andrews (1981) are reported for a variety of simulated theoretical output processes. In addition, each run is compared with various batch means procedures.