A multiobjective approach to evaluating and planning the allocation of inspection resources

A two-objective, zero-one programming model of an inspection allocation problem faced by a federal agency is solved by a two-stage decision support system. The decision support system allows decision-makers to explore the entire range of outputs from alternative investments. A two-stage decision support system, called the national model for inspection selection (NMIS), was developed to solve an inspection allocation problem faced by the Food and Drug Administration (FDA). The FDA in its regulation of the pharmaceutical industry must determine the ‘best’ level of inspection resources to allocate nationally and among its districts, while satisfying at the same time mandated requirements for inspection frequency. The problem is initially formulated here as a large-scale, single-objective, zero-one programming model. The authors decompose this initial formulation into a sequence of smaller two-objective district-level subproblems and one two-objective national problem. Algorithms were developed to solve both the district-level subproblems and the national problem for the entire range of outputs from possible alternative investments. The NMIS decision support system is comprised of a set of computer programs which performs these algorithms. NMIS provides FDA management with a powerful tool to assist them in developing and evaluating their decisions at various levels of limited resources.