Using quantiles in ranking and selection procedures

Jennifer Bekki, John Fowler, Gerald T. Mackulak, Barry L. Nelson

Research output: Chapter in Book/Report/Conference proceedingConference contribution

11 Citations (Scopus)

Abstract

A useful performance measure on which to compare manufacturing systems is a quantile of the cycle time distribution. Unfortunately, aside from order statistic estimates, which can require significant data storage, the distribution of quantile estimates has not been shown to be normally distributed, violating a common assumption amongst ranking-and-selection (R&S) procedures. To address this, we provide empirical evidence supporting an approach using the mean of a group of quantile estimates as the comparison measure. The approach is detailed and illustrated through experimentation on four M/M/1 queues in which the 0.9 cycle-time quantile is the performance measure. Results in terms of simulation effort and accuracy are reported and compared to results obtained using the macro-replications approach for inducing normality as well as to results obtained by applying R&S procedures to quantile estimates directly. The suggested procedure is shown to provide significant savings in simulation effort while sacrificing very little in accuracy.

Original languageEnglish (US)
Title of host publicationProceedings - Winter Simulation Conference
Pages1722-1728
Number of pages7
DOIs
StatePublished - 2007
Event2007 Winter Simulation Conference, WSC - Washington, DC, United States
Duration: Dec 9 2007Dec 12 2007

Other

Other2007 Winter Simulation Conference, WSC
CountryUnited States
CityWashington, DC
Period12/9/0712/12/07

Fingerprint

Macros
Statistics
Data storage equipment

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Bekki, J., Fowler, J., Mackulak, G. T., & Nelson, B. L. (2007). Using quantiles in ranking and selection procedures. In Proceedings - Winter Simulation Conference (pp. 1722-1728). [4419795] https://doi.org/10.1109/WSC.2007.4419795

Using quantiles in ranking and selection procedures. / Bekki, Jennifer; Fowler, John; Mackulak, Gerald T.; Nelson, Barry L.

Proceedings - Winter Simulation Conference. 2007. p. 1722-1728 4419795.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Bekki, J, Fowler, J, Mackulak, GT & Nelson, BL 2007, Using quantiles in ranking and selection procedures. in Proceedings - Winter Simulation Conference., 4419795, pp. 1722-1728, 2007 Winter Simulation Conference, WSC, Washington, DC, United States, 12/9/07. https://doi.org/10.1109/WSC.2007.4419795
Bekki J, Fowler J, Mackulak GT, Nelson BL. Using quantiles in ranking and selection procedures. In Proceedings - Winter Simulation Conference. 2007. p. 1722-1728. 4419795 https://doi.org/10.1109/WSC.2007.4419795
Bekki, Jennifer ; Fowler, John ; Mackulak, Gerald T. ; Nelson, Barry L. / Using quantiles in ranking and selection procedures. Proceedings - Winter Simulation Conference. 2007. pp. 1722-1728
@inproceedings{622aa7da0bec4b7fb3a1dd149e5f1b25,
title = "Using quantiles in ranking and selection procedures",
abstract = "A useful performance measure on which to compare manufacturing systems is a quantile of the cycle time distribution. Unfortunately, aside from order statistic estimates, which can require significant data storage, the distribution of quantile estimates has not been shown to be normally distributed, violating a common assumption amongst ranking-and-selection (R&S) procedures. To address this, we provide empirical evidence supporting an approach using the mean of a group of quantile estimates as the comparison measure. The approach is detailed and illustrated through experimentation on four M/M/1 queues in which the 0.9 cycle-time quantile is the performance measure. Results in terms of simulation effort and accuracy are reported and compared to results obtained using the macro-replications approach for inducing normality as well as to results obtained by applying R&S procedures to quantile estimates directly. The suggested procedure is shown to provide significant savings in simulation effort while sacrificing very little in accuracy.",
author = "Jennifer Bekki and John Fowler and Mackulak, {Gerald T.} and Nelson, {Barry L.}",
year = "2007",
doi = "10.1109/WSC.2007.4419795",
language = "English (US)",
isbn = "1424413060",
pages = "1722--1728",
booktitle = "Proceedings - Winter Simulation Conference",

}

TY - GEN

T1 - Using quantiles in ranking and selection procedures

AU - Bekki, Jennifer

AU - Fowler, John

AU - Mackulak, Gerald T.

AU - Nelson, Barry L.

PY - 2007

Y1 - 2007

N2 - A useful performance measure on which to compare manufacturing systems is a quantile of the cycle time distribution. Unfortunately, aside from order statistic estimates, which can require significant data storage, the distribution of quantile estimates has not been shown to be normally distributed, violating a common assumption amongst ranking-and-selection (R&S) procedures. To address this, we provide empirical evidence supporting an approach using the mean of a group of quantile estimates as the comparison measure. The approach is detailed and illustrated through experimentation on four M/M/1 queues in which the 0.9 cycle-time quantile is the performance measure. Results in terms of simulation effort and accuracy are reported and compared to results obtained using the macro-replications approach for inducing normality as well as to results obtained by applying R&S procedures to quantile estimates directly. The suggested procedure is shown to provide significant savings in simulation effort while sacrificing very little in accuracy.

AB - A useful performance measure on which to compare manufacturing systems is a quantile of the cycle time distribution. Unfortunately, aside from order statistic estimates, which can require significant data storage, the distribution of quantile estimates has not been shown to be normally distributed, violating a common assumption amongst ranking-and-selection (R&S) procedures. To address this, we provide empirical evidence supporting an approach using the mean of a group of quantile estimates as the comparison measure. The approach is detailed and illustrated through experimentation on four M/M/1 queues in which the 0.9 cycle-time quantile is the performance measure. Results in terms of simulation effort and accuracy are reported and compared to results obtained using the macro-replications approach for inducing normality as well as to results obtained by applying R&S procedures to quantile estimates directly. The suggested procedure is shown to provide significant savings in simulation effort while sacrificing very little in accuracy.

UR - http://www.scopus.com/inward/record.url?scp=49749141994&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=49749141994&partnerID=8YFLogxK

U2 - 10.1109/WSC.2007.4419795

DO - 10.1109/WSC.2007.4419795

M3 - Conference contribution

AN - SCOPUS:49749141994

SN - 1424413060

SN - 9781424413065

SP - 1722

EP - 1728

BT - Proceedings - Winter Simulation Conference

ER -