Variable selection for Gaussian process models in computer experiments

Crystal Linkletter, Derak Bingham, Nicholas Hengartner, David Higdon, Qian K. Ye

Research output: Contribution to journalArticle

70 Citations (Scopus)

Abstract

In many situations, simulation of complex phenomena requires a large number of inputs and is computationally expensive. Identifying the inputs that most impact the system so that these factors can be further investigated can be a critical step in the scientific endeavor. In computer experiments, it is common to use a Gaussian spatial process to model the output of the simulator. In this article we introduce a new, simple method for identifying active factors in computer screening experiments. The approach is Bayesian and only requires the generation of a new inert variable in the analysis; however, in the spirit of frequentist hypothesis testing, the posterior distribution of the inert factor is used as a reference distribution against which the importance of the experimental factors can be assessed. The methodology is demonstrated on an application in material science, a computer experiment from the literature, and simulated examples.

Original languageEnglish (US)
Pages (from-to)478-490
Number of pages13
JournalTechnometrics
Volume48
Issue number4
DOIs
StatePublished - Nov 2006

Fingerprint

Computer Experiments
Gaussian Model
Variable Selection
Gaussian Process
Process Model
Experiments
Materials science
Screening Experiment
Spatial Process
Materials Science
Screening
Hypothesis Testing
Simulators
Posterior distribution
Simulator
Testing
Methodology
Output
Simulation
Model

Keywords

  • Computer simulation
  • Latin hypercube
  • Random field
  • Screening
  • Spatial process

ASJC Scopus subject areas

  • Mathematics(all)
  • Statistics and Probability

Cite this

Variable selection for Gaussian process models in computer experiments. / Linkletter, Crystal; Bingham, Derak; Hengartner, Nicholas; Higdon, David; Ye, Qian K.

In: Technometrics, Vol. 48, No. 4, 11.2006, p. 478-490.

Research output: Contribution to journalArticle

Linkletter, C, Bingham, D, Hengartner, N, Higdon, D & Ye, QK 2006, 'Variable selection for Gaussian process models in computer experiments', Technometrics, vol. 48, no. 4, pp. 478-490. https://doi.org/10.1198/004017006000000228
Linkletter, Crystal ; Bingham, Derak ; Hengartner, Nicholas ; Higdon, David ; Ye, Qian K. / Variable selection for Gaussian process models in computer experiments. In: Technometrics. 2006 ; Vol. 48, No. 4. pp. 478-490.
@article{64bc7e4374514a22b6043bfe9889f5ae,
title = "Variable selection for Gaussian process models in computer experiments",
abstract = "In many situations, simulation of complex phenomena requires a large number of inputs and is computationally expensive. Identifying the inputs that most impact the system so that these factors can be further investigated can be a critical step in the scientific endeavor. In computer experiments, it is common to use a Gaussian spatial process to model the output of the simulator. In this article we introduce a new, simple method for identifying active factors in computer screening experiments. The approach is Bayesian and only requires the generation of a new inert variable in the analysis; however, in the spirit of frequentist hypothesis testing, the posterior distribution of the inert factor is used as a reference distribution against which the importance of the experimental factors can be assessed. The methodology is demonstrated on an application in material science, a computer experiment from the literature, and simulated examples.",
keywords = "Computer simulation, Latin hypercube, Random field, Screening, Spatial process",
author = "Crystal Linkletter and Derak Bingham and Nicholas Hengartner and David Higdon and Ye, {Qian K.}",
year = "2006",
month = "11",
doi = "10.1198/004017006000000228",
language = "English (US)",
volume = "48",
pages = "478--490",
journal = "Technometrics",
issn = "0040-1706",
publisher = "American Statistical Association",
number = "4",

}

TY - JOUR

T1 - Variable selection for Gaussian process models in computer experiments

AU - Linkletter, Crystal

AU - Bingham, Derak

AU - Hengartner, Nicholas

AU - Higdon, David

AU - Ye, Qian K.

PY - 2006/11

Y1 - 2006/11

N2 - In many situations, simulation of complex phenomena requires a large number of inputs and is computationally expensive. Identifying the inputs that most impact the system so that these factors can be further investigated can be a critical step in the scientific endeavor. In computer experiments, it is common to use a Gaussian spatial process to model the output of the simulator. In this article we introduce a new, simple method for identifying active factors in computer screening experiments. The approach is Bayesian and only requires the generation of a new inert variable in the analysis; however, in the spirit of frequentist hypothesis testing, the posterior distribution of the inert factor is used as a reference distribution against which the importance of the experimental factors can be assessed. The methodology is demonstrated on an application in material science, a computer experiment from the literature, and simulated examples.

AB - In many situations, simulation of complex phenomena requires a large number of inputs and is computationally expensive. Identifying the inputs that most impact the system so that these factors can be further investigated can be a critical step in the scientific endeavor. In computer experiments, it is common to use a Gaussian spatial process to model the output of the simulator. In this article we introduce a new, simple method for identifying active factors in computer screening experiments. The approach is Bayesian and only requires the generation of a new inert variable in the analysis; however, in the spirit of frequentist hypothesis testing, the posterior distribution of the inert factor is used as a reference distribution against which the importance of the experimental factors can be assessed. The methodology is demonstrated on an application in material science, a computer experiment from the literature, and simulated examples.

KW - Computer simulation

KW - Latin hypercube

KW - Random field

KW - Screening

KW - Spatial process

UR - http://www.scopus.com/inward/record.url?scp=33845260356&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33845260356&partnerID=8YFLogxK

U2 - 10.1198/004017006000000228

DO - 10.1198/004017006000000228

M3 - Article

VL - 48

SP - 478

EP - 490

JO - Technometrics

JF - Technometrics

SN - 0040-1706

IS - 4

ER -