A Modified Response Surface Methodology for Knowledge Discovery with Simulations

The response surface methodology (RSM) provides an iterative process for learning that involves the sequential use of experimental design, empirical model building, and analysis of the developed models. This work provides a modified response surface methodology (MRSM) that can be applied to more complex simulation studies. These problems involve a larger number of input variables, multiple measures of performance, and complex systems relationships. Furthermore, the MRSM considers simulation analyses that have multiple objectives. These may include: 1) determining near optimal solutions; 2) understanding tradeoffs; and 3) translating the findings into generalizations. In many applications, one may begin the analysis with relatively little understanding of the variable relationships in the systems under study. The MRSM capitalizes on the underlying learning philosophy of the traditional RSM while benefiting from other knowledge discovery concepts and data mining techniques. We describe the general concepts behind the MRSM and a sequential procedure for analysis of a broad range of simulation applications. To show the steps of the MRSM, we provide examples based on an application involving combat agent-based simulation models.