A refined selection method for project portfolio optimization considering project interactions

This paper proposes two innovative approaches to address the two challenges of project portfolio selection: (i) Determining how project interactions influence the final values of project portfolios, and (ii) selecting the best solution from non-dominated project portfolios. First, based on the dependence relationship between projects and technologies, we construct a project co-utilization network indicating project interactions using the co-citation network method. We also consider project interaction as a regularization item of the value function to indicate the influence of project interactions on the value of project portfolios. Second, we propose a refined selection strategy. All the projects are ranked in the order of their appearance in the ranked association rules. Project portfolios having no highly ranked projects are deleted successively until only one of them remains. Finally, a case of multi-objective project portfolio selection is studied and an optimal project portfolio recommended. A comparison shows the availability and advantages of the proposed approaches. (C) 2019 Elsevier Ltd. All rights reserved.