Two-stage sensitivity-based group screening in computer experiments
Two-stage sensitivity-based group screening in computer experiments
Sophisticated computer codes that implement mathematical models of physical processes can involve large numbers of inputs, and screening to determine the most active inputs is critical for understanding the input- output relationship. This article presents a new two-stage group screening methodology for identifying active inputs. In Stage 1, groups of inputs showing low activity are screened out; in Stage 2, individual inputs from the active groups are identified. Inputs are evaluated through their estimated total (effect) sensitivity indices (TSIs), which are compared with a benchmark null TSI distribution created from added low noise inputs. Examples show that, compared with other procedures, the proposed method provides more consistent and accurate results for high-dimensional screening. Additional details and computer code are provided in supplementary materials available online.
active factor, experimental design, gaussian process, latin hypercube design, low-impact input, total sensitivity index
376-387
Moon, Hyejung
486b8369-2c4c-49e8-8239-5e1d25993e3f
Dean, Angela M.
9c90540a-cdf4-44ce-9d34-6b7b495a1ea3
Santner, Thomas J.
25c0e9d6-daa2-4757-bee2-a714805eb67c
28 November 2012
Moon, Hyejung
486b8369-2c4c-49e8-8239-5e1d25993e3f
Dean, Angela M.
9c90540a-cdf4-44ce-9d34-6b7b495a1ea3
Santner, Thomas J.
25c0e9d6-daa2-4757-bee2-a714805eb67c
Moon, Hyejung, Dean, Angela M. and Santner, Thomas J.
(2012)
Two-stage sensitivity-based group screening in computer experiments.
Technometrics, 54 (4), .
(doi:10.1080/00401706.2012.725994).
Abstract
Sophisticated computer codes that implement mathematical models of physical processes can involve large numbers of inputs, and screening to determine the most active inputs is critical for understanding the input- output relationship. This article presents a new two-stage group screening methodology for identifying active inputs. In Stage 1, groups of inputs showing low activity are screened out; in Stage 2, individual inputs from the active groups are identified. Inputs are evaluated through their estimated total (effect) sensitivity indices (TSIs), which are compared with a benchmark null TSI distribution created from added low noise inputs. Examples show that, compared with other procedures, the proposed method provides more consistent and accurate results for high-dimensional screening. Additional details and computer code are provided in supplementary materials available online.
Text
MoonDeanSantner-Technometrics2012.pdf
- Author's Original
Restricted to Repository staff only
Request a copy
More information
Published date: 28 November 2012
Keywords:
active factor, experimental design, gaussian process, latin hypercube design, low-impact input, total sensitivity index
Organisations:
Mathematical Sciences
Identifiers
Local EPrints ID: 349680
URI: http://eprints.soton.ac.uk/id/eprint/349680
ISSN: 0040-1706
PURE UUID: 2a8f733e-882c-43a5-b54b-1f0340a2f4b4
Catalogue record
Date deposited: 07 May 2013 08:44
Last modified: 27 Oct 2023 01:58
Export record
Altmetrics
Contributors
Author:
Hyejung Moon
Author:
Angela M. Dean
Author:
Thomas J. Santner
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics