|
Activity Number:
|
602
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #304486 |
|
Title:
|
Resampling Methods for Queue Inference
|
|
Author(s):
|
Harry Ma*+ and Thomas Willemain
|
|
Companies:
|
GE Global Research and Rensselaer Polytechnic Institute
|
|
Address:
|
, , ,
|
|
Keywords:
|
bootstrap ; queue ; simulation ; resampling
|
|
Abstract:
|
Given a computationally expensive discrete event simulation model, it is desirable to make reliable inferences from a single replication. Bootstrap resampling provides an approach to this problem. The conventional approach to bootstrapping a time series uses either the moving block bootstrap (MBB) or the stationary bootstrap (SB). We compare the performance of the MBB and SB for estimating the standard error of the mean delay in an M/M/1 queue against two alternatives: the regenerative bootstrap and a new method that resamples changes.
|
