ABSTRACT
This chapter explores a number of distinct topics. First, the analysis and interpretation of simulation results are discussed with the use of the Advanced Sensitivity Analysis tool within @RISK—these results allow a more granular and probing examination of simulation results, and allow us to understand which are the key risk drivers and further understand the shapes of output distributions. In addition, stress concerns the analysis of possible, but extreme, outcomes—that is, examining the tails of output probability distributions.
The second section explores different techniques that can be used when building simulation models. This includes frequency-severity problems that combine discrete probability distributions (the “frequency” of events) with continuous probability distributions (the “severity” of the consequences). Also examined are multimodal input distributions or other distributions that are not adequately modeled by parametric probability distribution.
The final section of this chapter is devoted to Bayesian analysis, where input data into a simulation model is combined with existing “knowledge”, as steps in an ongoing learning process. Bayesian analysis is a large and rapidly expanding field (it's not new, though—Thomas Bayes died in 1761), and this section gives the reader an idea of how Bayesian analysis can be used in a simulation model.
