Model-robust supersaturated and partially supersaturated designs

Bradley A. Jones, William Li, Christopher J. Nachtsheim, Qian K. Ye

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

Supersaturated designs are an increasingly popular tool for screening factors in the presence of effect sparsity. The advantage of this class of designs over resolution III factorial designs or Plackett-Burman designs is that n, the number of runs, can be substantially smaller than the number of factors, m. A limitation associated with most supersaturated designs produced thus far is that the capability of these designs for estimating g active effects has not been discussed. In addition to exploring this capability, we develop a new class of model-robust supersaturated designs that, for a given n and m, maximizes the number g of active effects that can be estimated simultaneously. The capabilities of model-robust supersaturated designs for model discrimination are assessed using a model-discrimination criterion, the subspace angle. Finally, we introduce the class of partially supersaturated designs, intended for use when we require a specific subset of m1 core factors to be estimable, and the sparsity of effects principle applies to the remaining (m - m1) factors.

Original languageEnglish (US)
Pages (from-to)45-53
Number of pages9
JournalJournal of Statistical Planning and Inference
Volume139
Issue number1
DOIs
StatePublished - Jan 2009

Fingerprint

Supersaturated Design
Model Discrimination
Robust Design
Effect Sparsity
Plackett-Burman Design
Model
Factorial Design
Sparsity
Screening
Maximise
Subspace
Angle
Subset
Factors
Class

Keywords

  • Estimation capacity
  • Exchange algorithm
  • Model-robust design
  • Optimal design
  • Partially supersaturated design
  • Supersaturated design

ASJC Scopus subject areas

  • Statistics, Probability and Uncertainty
  • Applied Mathematics
  • Statistics and Probability

Cite this

Model-robust supersaturated and partially supersaturated designs. / Jones, Bradley A.; Li, William; Nachtsheim, Christopher J.; Ye, Qian K.

In: Journal of Statistical Planning and Inference, Vol. 139, No. 1, 01.2009, p. 45-53.

Research output: Contribution to journalArticle

Jones, Bradley A. ; Li, William ; Nachtsheim, Christopher J. ; Ye, Qian K. / Model-robust supersaturated and partially supersaturated designs. In: Journal of Statistical Planning and Inference. 2009 ; Vol. 139, No. 1. pp. 45-53.
@article{f69ac15cbfa9409db020d2cad26cbb8a,
title = "Model-robust supersaturated and partially supersaturated designs",
abstract = "Supersaturated designs are an increasingly popular tool for screening factors in the presence of effect sparsity. The advantage of this class of designs over resolution III factorial designs or Plackett-Burman designs is that n, the number of runs, can be substantially smaller than the number of factors, m. A limitation associated with most supersaturated designs produced thus far is that the capability of these designs for estimating g active effects has not been discussed. In addition to exploring this capability, we develop a new class of model-robust supersaturated designs that, for a given n and m, maximizes the number g of active effects that can be estimated simultaneously. The capabilities of model-robust supersaturated designs for model discrimination are assessed using a model-discrimination criterion, the subspace angle. Finally, we introduce the class of partially supersaturated designs, intended for use when we require a specific subset of m1 core factors to be estimable, and the sparsity of effects principle applies to the remaining (m - m1) factors.",
keywords = "Estimation capacity, Exchange algorithm, Model-robust design, Optimal design, Partially supersaturated design, Supersaturated design",
author = "Jones, {Bradley A.} and William Li and Nachtsheim, {Christopher J.} and Ye, {Qian K.}",
year = "2009",
month = "1",
doi = "10.1016/j.jspi.2008.05.015",
language = "English (US)",
volume = "139",
pages = "45--53",
journal = "Journal of Statistical Planning and Inference",
issn = "0378-3758",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Model-robust supersaturated and partially supersaturated designs

AU - Jones, Bradley A.

AU - Li, William

AU - Nachtsheim, Christopher J.

AU - Ye, Qian K.

PY - 2009/1

Y1 - 2009/1

N2 - Supersaturated designs are an increasingly popular tool for screening factors in the presence of effect sparsity. The advantage of this class of designs over resolution III factorial designs or Plackett-Burman designs is that n, the number of runs, can be substantially smaller than the number of factors, m. A limitation associated with most supersaturated designs produced thus far is that the capability of these designs for estimating g active effects has not been discussed. In addition to exploring this capability, we develop a new class of model-robust supersaturated designs that, for a given n and m, maximizes the number g of active effects that can be estimated simultaneously. The capabilities of model-robust supersaturated designs for model discrimination are assessed using a model-discrimination criterion, the subspace angle. Finally, we introduce the class of partially supersaturated designs, intended for use when we require a specific subset of m1 core factors to be estimable, and the sparsity of effects principle applies to the remaining (m - m1) factors.

AB - Supersaturated designs are an increasingly popular tool for screening factors in the presence of effect sparsity. The advantage of this class of designs over resolution III factorial designs or Plackett-Burman designs is that n, the number of runs, can be substantially smaller than the number of factors, m. A limitation associated with most supersaturated designs produced thus far is that the capability of these designs for estimating g active effects has not been discussed. In addition to exploring this capability, we develop a new class of model-robust supersaturated designs that, for a given n and m, maximizes the number g of active effects that can be estimated simultaneously. The capabilities of model-robust supersaturated designs for model discrimination are assessed using a model-discrimination criterion, the subspace angle. Finally, we introduce the class of partially supersaturated designs, intended for use when we require a specific subset of m1 core factors to be estimable, and the sparsity of effects principle applies to the remaining (m - m1) factors.

KW - Estimation capacity

KW - Exchange algorithm

KW - Model-robust design

KW - Optimal design

KW - Partially supersaturated design

KW - Supersaturated design

UR - http://www.scopus.com/inward/record.url?scp=52749094527&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=52749094527&partnerID=8YFLogxK

U2 - 10.1016/j.jspi.2008.05.015

DO - 10.1016/j.jspi.2008.05.015

M3 - Article

AN - SCOPUS:52749094527

VL - 139

SP - 45

EP - 53

JO - Journal of Statistical Planning and Inference

JF - Journal of Statistical Planning and Inference

SN - 0378-3758

IS - 1

ER -