ABSTRACT
With either approach, stochastic or deterministic, variation in outputs (whether due to randomness, input manipulation, or both) is the basis for statistical analysis of outputs. This chapter addresses statistical analysis of outputs for stochastic models, with the understanding that some of the analytical techniques can also be applied to deterministic models by varying input values systematically. First, simulation is a statistical sampling experiment in which models convert stochastic inputs into statistical data output. Because simulation is a sampling process, variable estimates are subject to sampling choice and sample size. Since random samples from probability distributions are used as simulation inputs, the outputs (estimates) for each run are themselves random variables that may have large variances. It is important to distinguish between system classes and states for performing output analysis. The analysis method may differ depending on whether the system is a terminating system or non-terminating system; and whether the system is in a transient state or steady state.
