全局敏感性分析可用于确定和排列设计目标和约束的变量重要性（敏感性），其中对解空间进行采样，通常以逐步方式采用线性回归模型。变量的相对重要性可以通过三个指标来检验:变量进入线性回归模型的顺序；标准化回归系数或其秩变换系数的绝对值；R2的大小变化（决定系数）归因于每一步的附加变量。然而，由逐步回归构建的线性回归模型的稳健性与程序选项的选择有关，例如:。 g、 样本集和数据公式。不同的程序选项可能导致不同的线性回归模型，从而影响变量的全局敏感性。因此，本文研究了逐步回归的程序选项在多大程度上会影响变量的指示——能源需求、资本成本和解决方案不可行性的三个不同敏感性指标测量的全球敏感性，当使用随机生成的样本和多目标优化过程（基于NSGA-II）开始时获得的偏差解时。它的结论是，无论选择什么程序选项，最重要的变量总是排在第一位，但最好同时采用这两个条目- 变量顺序及其标准化（秩）回归系数或R2变化的贡献，为设计目标和约束提供变量重要性的稳健顺序。此外，当样本量较小时，重新生成的灵敏度分析样本集可以避免误导变量的重要性，特别是对于中间变量的排序。最后，为了提高计算效率，本文得出结论，从多目标优化中获得的前100个解可用于进行全局灵敏度分析，以确定设计目标的重要变量。引文:ASHRAE论文CD: 2014年ASHRAE年会，华盛顿州西雅图

Global sensitivity analysis can be used to identify and rank variables importance (sensitivities) for design objectives and constraints, wherethe solution space is sampled and a linear regression model is normally adopted in the stepwise manner. The relative importance of variables can beexamined by three indicators: the order of variables entry into the linear regression model; the absolute values of the standardized regressioncoefficients or their rank transformation coefficients; and the size of the R2 changes (coefficient of determination) attributable to additional variablesat each step. However, the robustness of the linear regression model constructed from a stepwise regression is related to the choice of procedureoptions, e.g. the set of samples and data formulation. Different procedure options could lead to different linear regression models, and thereforeinfluence the indication of variables global sensitivities. Thus, this paper investigates the extent to which the procedure options of a stepwiseregression can influence the indication of variables global sensitivities, measured by three different sensitivity indicators, for energy demand, capitalcosts and solution infeasibility, when using both the randomly generated samples and the biased solutions obtained at the start of a multi-objectiveoptimization process (based on NSGA-II). It concludes that the most important variables are always ranked on the top no matter the choice ofprocedure options, but it is better to adopt both the entry-orders of variables and their standardized (rank) regression coefficients or thecontributions to R2 changes, to provide robust orderings of variables importance, for design objectives and constraints. Moreover, when the samplesize is smaller, re-generated separate set of samples for sensitivity analysis can avoid misleading variables importance, especially for the variablesranked in the middle. Finally, to improve computational efficiency, this paper concludes that the first 100 solutions obtained from a multi-objectiveoptimization can be used to perform global sensitivity analysis, to identify the important variables for design objectives.