Graduating sample data using generalized Weibull functions

Most activities, or events, in discrete event simulation models are probabilistic in nature and describable by density functions. Consequently, persons developing simulation models are frequently required to select and fit density functions to sample data and, correspondingly, to determine the process, or random-event, generator.

This paper seeks to establish that the graduation process involved with simulation modeling may be expedited: generalized Weibull functions numerically fitted by nonlinear least-squares or maximum-likelihood procedures are comparable in performance to the fits of more traditional functions. The use of a single function simplifies the ambiguous process of function selection, while offering, in addition, the important advantages of a closed-form inverse process generator.