Finding compromise solutions in project portfolio selection with multiple experts by inverse optimization

This paper deals with project portfolio selection evaluated by multiple experts. The problem consists of selecting a subset of projects that satisfies a set of constraints and represents a compromise among the group of experts. It can be modeled as a multi-objective combinatorial optimization problem and solved by two procedures based on inverse optimization. It requires to find a minimal adjustment of the expert's evaluations such that a portfolio becomes ideal in the objective space. Several distance functions are considered to define a measure of the adjustment. The two procedures are applied to randomly generated instances of the knapsack problem and computational results are reported. Finally, two illustrative examples are analyzed and several theoretical properties are proved.