Bayesian Analysis for Transportation Risk

摘要

High profile terrorist attacks featured the transportation system as the target (e.g. London in 2005, Madrid in 2004, and Tokyo in 1995) and the weapon (e.g. USA in 2001). Completely securing the transportation system against exploitation in these capacities and as a means to reach other targets is an expensive challenge due to the necessary' openness of the system for the efficient movement of people and goods. Thus, the need exists for a comprehensive, dynamic risk assessment, which must account for the probability of various events as well as thefr consequences. While specific attacks may be infrequent, their probabilities should not necessarily be considered zero. Experts can estimate these low probabilities based on their "degree of belief about the scenario. Since these probabilities are subjective, they can be updated as additional evidence becomes available. This paper demonstrates how Bayesian analysis can be used to update attack scenario probabilities after receiving new information. Two different evidence specificities are analyzed here, but the method is not limited to this number. The updating process depends on the total probability of receiving information of a given specificity and the conditional probability that information is received given that the scenario is actually selected. Numerical experiments demonstrate that false information may reduce the probability of a scenario and has the power to influence the defender's risk assessment even in the face of intelligence information. Furthermore, false information may lead the defender to expend valuable, limited resources to protect a site that is not the planned target.