Bachhofer-Gupta-Sobel Memorial Symposium on
Ranking and Selection and Multiple Comparison Procedures Methodologies
|
Dave
Goldsman
is a Professor in the School of Industrial and Systems Engineering
at Georgia Tech. His research is concerned with the application
of statistical techniques for use in the analysis of simulation
processes. He is particularly interested in combining simulation
output analysis with techniques of statistical ranking and
selection, and he investigates such questions as: What kind of
performance can a practitioner expect from a particular simulated
system? Or which of a number of competing systems can be
characterized as the ``best''? He has publishes primarily in such
archival journals as IIE Transactions, Management
Science, Naval Research Logistics, Operations Research, and
Operations Research Letters. Dave is also an active member and
participant in INFORMS, IIE, and Winter Simulation Conference
activities; in particular, he was the WSC's Program Chair in 1995
and is presently on its Board of Directors. He was recently
Editor for the Simulation Department of IIE Transactions
and is currently an Associate Editor for Sequential Analysis.
|
Some Contributions of Ranking and
Selection to Computer Simulation
Over the last 30 years, the seminal work of
Bechhofer, Gupta, and Sobel (BG S) has led to a array of advances in
the field of computer simulation, particularly in the area of
simulation output analysis. In this talk, we describe three modern
procedures for deciding which of a number of competing simulated
systems is the best. These three procedures trace their roots directly
to earlier procedures due to BG S. Each of the new procedures takes
advantage of certain inherent characteristics of simulation, e.g., the
abilities to induce correlation between competing systems, quickly
produce and analyze observations sequentially, and efficiently
re-sample observations.
|