Abstract for:Addressing parameter uncertainty in SD models with fit-to-history and Monte-Carlo sensitivity methods

We present a practical guide, including a step-by-step flowchart, for establishing confidence intervals for key model outcomes in the face of uncertain parameters. The process starts with Powell optimization to find a set of uncertain parameters (the OPS) that minimize the model fitness error relative to available reference behavior data. The optimization process also helps to refine assumed parameter uncertainty ranges.  Next, Markov Chain Monte Carlo or conventional Monte Carlo randomization is used to create a sample of parameter sets that fit the reference behavior data nearly as well as the OPS.  Using MC method, the entire parameter space is explored broadly (millions of runs), and the results sorted to select qualifying parameter sets (QPS) based on goodness-of-fit criteria.  The statistical properties of the QPS parameter distributions are analyzed to ensure their centrality relative to the uncertainty ranges.  The full set of QPS outputs are displayed as sensitivity graphs overlaid with reference behavior data.  Finally, alternative scenarios are evaluated via the OPS and QPS, and confidence intervals determined for key model outcomes.  The method is demonstrated with a non-trivial model, and a narrative template is provided showing how such analyses could be described to interested parties such as policy makers.