Optimizing airline passenger prescreening systems with Bayesian decision models

The Transportation Security Agency provides airline security in the United States using a variety of measures including a computer based passenger prescreening system. This paper develops Bayesian decision models of two prescreening systems: one that places ticketed passengers into two classifications (fly and no-fly), and a three classification system that includes potential flight. Using a parameterized cost structure, and the expected monetary value decision criteria, this paper develops optimal levels of undesirable personal characteristics that should place people into the various categories. The models are explored from both the government perspective and the passenger's perspective.