Quantitative Metrics in MS VVA

摘要

The theoretic approach to consider a simulation model binary as valid or invalid for a particular intended purpose is not satisfactory in real practice. The question rather is, which of the requirements that the simulation model needs to pass to be considered as valid for the intended purpose are perceived as passed, and how certain can one be concerning this perception. Based on this perception and the uncertainty associated with it, the decision maker can analyze the risk of using the simulation executable model for the original intended purpose or any of its (degraded) derivates. The paper proposes an approach how to formally specify a hierarchy of acceptability criteria, to quantify the uncertainty associated with the perception that each individual acceptability criterion was passed or failed, and to estimate the risk associated with accepting the simulation executable model for a desired intended purpose. Concepts from propositional logic, hierarchical breakdown structures, possibility theory, stochastic simulation, and decision theory are combined into one integrated approach.