A multiobjective optimization model for project selection with probabilistic considerations

When faced with limited resources, project managers must determine which projects to fund at what levels from a pool of potential ones. This problem of project selection is inherently multiobjective since various factors, such as the available budget, the chance of success, and the efficient allocation of the project team, must be considered simultaneously. The uncertainty of the data at the time decisions are made further complicates project selection. In this paper, a multiobjective, integer-constrained optimization model with competing objectives for project selection is formulated using probability distributions to describe costs. The objectives correspond to important project criteria, such as: rank (value), managerial labor needed, and average cost. The subjective rank is determined via the Analytic Hierarchy Process. The model is applied to a data set from a US government agency that involves 84 separate projects. The results indicate improved budgetary efficiency compared to the actual project selection, thus supporting use of the model for public sector project selection. The model is unique since it integrates multiobjective optimization, Monte Carlo simulation, and the Analytic Hierarchy Process.