On coverage probability, independence and normality in batch means algorithms

Nabil R. Adam, Steven W. Klein

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

We consider the batching approach to interval estimation of the mean of a stochastic process based upon simulation output. In particular, using batch means processes generated from underlying AR(1) processes and M/M/1 queueing systems, we study the effects of undetected autocorrelation on both the probability of coverage and the performance of the Shapiro-Wilk test of normality. We also conduct power comparisons of two tests used to detect first-order autocorrelation in batch means processes: the von Neumann ratio test and a rank version thereof proposed by Bartels (1982). Our results indicate that a large number of batches (perhaps at least 100) be used when testing for the presence of autocorrelation.

Original languageEnglish (US)
Pages (from-to)206-215
Number of pages10
JournalEuropean Journal of Operational Research
Volume43
Issue number2
DOIs
StatePublished - Nov 27 1989

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Modeling and Simulation
  • Management Science and Operations Research
  • Information Systems and Management

Keywords

  • Simulation
  • batch means
  • output analysis

Fingerprint Dive into the research topics of 'On coverage probability, independence and normality in batch means algorithms'. Together they form a unique fingerprint.

Cite this