Probabilistic Risk Analysis Versus Decision Analysis: Similarities, Differences and Illustrations

摘要: The methods of engineering probabilistic risk analysis and expected-utility decision share a common core: model occurrences uncertain events. This is based on systems the identification an exhaustive mutually exclusive set scenarios, their probabilities consequences. Both rely assumption rationality use Bayesian probability, both assume separation probability assessments preferences among scenarios’ outcomes. major differences are rooted in nature framing problems that they address. A often performed before decisions have been fully defined, one its objectives then to identify characterize mitigation options. Furthermore, at time analysis, maker who will eventually results unknown. Therefore, definition as degree belief has be adapted, for instance, by assuming implicit delegation user’s judgment analyst experts, which requires special care presentation results. Also, single system (e.g., aircraft) unit or operation takeoff landing cycle) when reality, may intended support management concern unknown number similar unspecified units. multiplicity implications treatment second-level uncertainties (about failure probabilities) need display these In this paper, two classical definitions (Bayesian frequentist) discussed, focusing relevance facing aleatory (randomness) well epistemic (limited knowledge about fundamental phenomenon interest). briefly described, along with similarities differences. Two illustrations presented: 1990, losing NASA orbiter crew due tiles thermal protection system, method assessment terrorist attack United States given frame, available intelligence information (a 2002 study). latter involves simple game involving alternating moves terrorists US using rational descriptive mode. main conclusion whereas role represent faithfully beliefs known order preferred alternative, needs scrupulous presenting assumptions, sources data processing allow future makers exercise own judgments