Revolutionize Simulation Experiments with Low-Order Regression and Kriging Metamodels
The tutorial discusses how to design and analyze simulation experiments for various purposes like validation, prediction, and optimization. It focuses on using low-order polynomial regression and Kriging metamodels to guide the experiment design. Before applying these metamodels, the inputs of the simulation model should be screened using sequential bifurcation. Optimization can be done using response surface methodology or Kriging models. "Robust" optimization considers uncertainty in simulation inputs. The tutorial also references earlier WSC papers.