Design of Experiment in Global Sensitivity Analysis Based on ANOVA High-Dimensional Model Representation

被引：21

作者：

Wang, Xiaodi ^{[1
]}

Tang, Yincai ^{[1
]}

Chen, Xueping ^{[2
]}

Zhang, Yingshan ^{[1
]}

机构：

[1] E China Normal Univ, Sch Finance & Stat, Shanghai 200241, Peoples R China

[2] Jiangsu Teachers Univ Technol, Dept Math, Changzhou, Peoples R China

来源：

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION | 2010年 / 39卷 / 06期

关键词：

Design of experiment; Global sensitivity analysis; Global sensitivity indices; Monte Carlo algorithm; Orthogonal decomposition;

D O I：

10.1080/03610918.2010.484122

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

070103 [概率论与数理统计]; 140311 [社会设计与社会创新];

摘要：

This article reviews global sensitivity analysis based on ANOVA high-dimensional model representation. To overcome the computational difficulties and explore the use of design of experiment (DOE) in global sensitivity analysis, two methods are presented. If the form of the objective function f is known, we use DOE to estimate the global sensitivity indices instead of Monte Carlo simulation. Otherwise, we use the observed values of the experiments to do global sensitivity analysis. These methods are easy to implement and can reduce the computational cost. An example is given to show the feasibility of replacing Monte Carlo (MC) or quasi-Monte Carlo (quasi-MC) simulation by design of experiment.

引用

页码：1183 / 1195

页数：13

共 22 条

[1]

[Anonymous], J E CHINA NORMAL U N

[2]

IMPACT OF HETEROGENEOUS REACTIONS ON STRATOSPHERIC CHEMISTRY OF THE ARCTIC [J].