Application of the mixed-effects model for determining optimal input variables in simulation analysis

Choice of desirable input variables is an important consideration in simulation models used for analyzing behaviors and investigating characteristics of a certain process or system. Usually, the exact information of the input variables needed for simulation analysis could not be known a priori, and so analysts typically use summary data information or assume a certain set of input variables which might be plausible for the simulation model and analysis. However, sometimes the initial choice of the input variables for a simulation model is incorrect. This can result in considerably unnecessary long simulation time or an undesirable objective function value which might be far from the real one. Thus, in this paper, we investigate the feasibility of using the mixed-effects model in determining appropriate input variables for the simulation and evaluate the quality of the optimal solutions obtained from applying this method with a case study.