روش های هموارسازی پیش بینی کننده چندگانه برای تجزیه و تحلیل حساسیت: نتایج نمونه
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|25949||2008||23 صفحه PDF||سفارش دهید||13950 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Reliability Engineering & System Safety, Volume 93, Issue 1, January 2008, Pages 55–77
The use of multiple predictor smoothing methods in sampling-based sensitivity analyses of complex models is investigated. Specifically, sensitivity analysis procedures based on smoothing methods employing the stepwise application of the following nonparametric regression techniques are described in the first part of this presentation: (i) locally weighted regression (LOESS), (ii) additive models, (iii) projection pursuit regression, and (iv) recursive partitioning regression. In this, the second and concluding part of the presentation, the indicated procedures are illustrated with both simple test problems and results from a performance assessment for a radioactive waste disposal facility (i.e., the Waste Isolation Pilot Plant). As shown by the example illustrations, the use of smoothing procedures based on nonparametric regression techniques can yield more informative sensitivity analysis results than can be obtained with more traditional sensitivity analysis procedures based on linear regression, rank regression or quadratic regression when nonlinear relationships between model inputs and model predictions are present.
The first part of this presentation  reviews parametric and nonparametric regression procedures for use in sampling-based sensitivity analyses. Specifically, the following parametric regression procedures are introduced and briefly described in Section 2 of Ref. : (i) linear regression (LIN_REG), (ii) rank regression (RANK_REG), and (iii) quadratic regression (QUAD_REG). Further, the following nonparametric regression procedures are introduced and briefly described in Section 3.3 of Ref. : (i) locally weighted regression (LOESS), (ii) additive models (GAMs), (iii) projection pursuit regression (PP_REG), and (iv) recursive partitioning regression (RP_REG). In addition, algorithms for the stepwise implementation of these procedures in the R language as part of a sensitivity analysis are described in Section 4 of Ref. . The efficacy of the various methods described in Ref.  as procedures for sensitivity analysis is now investigated with both analytic test model data and real data. The analytic test models were assembled as part of a review volume on sensitivity analysis  and . The real data comes from a performance assessment for the Waste Isolation Pilot Plant (WIPP)  and . The methods are compared on the basis of fidelity to the data, overfitting of the data, and reproducibility. The presentation is organized as follows. First, certain quantities used in assessing the efficacy of the various sensitivity analysis procedures are introduced (Section 2). Then, the results obtained with the analytic test models are presented (Section 3). Next, the results obtained with the data from the WIPP performance assessment are presented (Section 4). The presentation then ends with a summary of observations and insights (Section 5).