Using bivariate Bezier distributions to model simulation input processes

We describe a graphical, interactive technique for modeling bivariate simulation input processes using a distribution family based on Bezier curves and surfaces. This family has an open-ended parameterization and is capable of accurately representing an unlimited variety of shapes for marginal distributions together with many common types of bivariate stochastic dependence. Our input-modeling technique is implemented in a Windows-based software system called PRIME-PRobabilistic Input Modeling Environment. Several examples illustrate the application of PRIME to subjective and data-driven estimation of bivariate distributions representing simulation inputs.