Factor Screening in Simulation: Evaluation of Two Strategies Based on Random Balance Sampling
评估了两种基于随机平衡抽样的因子筛选策略，适用于输入变量多而模拟运行次数有限的情况，帮助识别重要变量，减少计算成本。
In the study of large, complex computer simulation models the user is often overwhelmed by the vast number of input variables. Moreover, he or she is usually confused about how to make an effective analysis of the model without performing an excessive number of runs, which tend to be costly and time consuming. Factor screening methods, which attempt to identify the more important variables, can be extremely useful in the study of such models. This paper presents and evaluates two screening strategies based upon random balance sampling. Both strategies are applicable when there are more variables to be screened than there are available screening runs. The results provide guidance in using these strategies in particular screening applications.