Orthogonal Arrays for the Estimation of Global Sensitivity Indices Based on ANOVA High-Dimensional Model Representation

We present an efficient new approach to estimate global sensitivity indices based on ANOVA high-dimensional model representation. The method makes use of the properties of orthogonal arrays and extend the results of Wang et al. ( 2010 ) to a more general situation. A theorem is given to illustrate t...

Full description

Saved in:
Bibliographic Details
Published inCommunications in statistics. Simulation and computation Vol. 40; no. 9; pp. 1324 - 1341
Main Authors Wang, Xiaodi, Tang, Yincai, Zhang, Yingshan
Format Journal Article
LanguageEnglish
Published Colchester Taylor & Francis Group 01.10.2011
Taylor & Francis
Taylor & Francis Ltd
Subjects
62-XX
Analysis of variance
Calculus of variations and optimal control
Combinatorics
Combinatorics. Ordered structures
Computation
Computer simulation
Designs and configurations
Estimates
Exact sciences and technology
Global sensitivity indices
Linear inference, regression
Mathematical analysis
Mathematical models
Mathematics
Monte Carlo simulation
Numerical analysis
Numerical analysis. Scientific computation
Numerical methods in probability and statistics
Orthogonal array
Orthogonal arrays
Probability and statistics
Quasi-Monte Carlo
Representations
Sampling
Sciences and techniques of general use
Statistics
Studies
Monte Carlo method
62-XX
Statistical distribution
Covariance analysis
Orthogonal array
Estimation
Global sensitivity indices
Stochastic method
Variance analysis
Statistical method
Statistical regression
Numerical analysis
Estimator efficiency
Numerical simulation
Approximation theory
Sampling
Quasi-Monte Carlo
Monte Carlo simulation
Online AccessGet full text

Cover

More Information
Summary:We present an efficient new approach to estimate global sensitivity indices based on ANOVA high-dimensional model representation. The method makes use of the properties of orthogonal arrays and extend the results of Wang et al. ( 2010 ) to a more general situation. A theorem is given to illustrate that both the non influential and significant sensitivity indices are well estimated. This new method offers higher sampling efficiency than those of Sobol and Saltelli. We test its performance on two different models, which are widely used in engineering and statistical areas.
Bibliography:ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
ISSN:0361-0918
1532-4141
DOI:10.1080/03610918.2011.575500