In this paper, we explain the important role of simulation input modeling in a successful simulation study. Two pitfalls in simulation input modeling are then presented and we explain how any analyst, regardless of their knowledge of statistics, can easily avoid these pitfalls through the use of ExpertFit, the Windows-based successor to the UniFit II input modeling package. We use a set of real-world system data to demonstrate how the package automatically specifies, evaluates, and ranks candidate probability distributions, and then assists an analyst in deciding whether the 'best' candidate probability distribution provides an adequate representation of the data. If no candidate probability distribution provides an adequate fit, then ExpertFit can define an empirical distribution function. In either case, the probability distribution can be automatically expressed in the analyst's simulation software. We then consider the general case of selecting a probability distribution in the absence of data. As an example, we show how ExpertFit can be used to create busy-time and downtime models for machines that are subject to random breakdowns.
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=293243.