Estimating relationships in simulation models using regression: An application to military retirement costing

Abstract

This article demonstrates a procedure for using regression analysis to develop estimating equations that simply and quickly predict the effects on system performance of changes in parametric values of complex interrelation- ships imbedded in a full mathematical simulation model. Though a analytic solution to the relationships may be possible, the time and other resources necessary to generate such a solution may exceed those available. The resulting estimating equations facilitate understanding of the system being simulated, enable users to more easily conduct sensitivity analysis and answer what-if questions, and assist the selling of the simulation results to potential users. The procedure includes criteria for selection of: variables to be analyzed, the sensitivity range, the value increments, and the functional form. The example utilized is a simulation model developed for estimating future military retirement costs. (Author) ; Prepared for: The Office of the Assistant Secretary of Defense (MRA&L), Washington DC ; http://archive.org/details/estimatingrelati00eusk ; Funding for this research was provided by the Office of the Assistant Secretary of Defense (Manpower, Installations, and Logistics). ; NA