A model to reduce the risk of project selection utilizing data envelopment analysis

Abstract

Purpose Project selection management is a matter of challenge for project-oriented organizations, particularly, if the decision-makers are confronted with limited resources. One of the main concerns is selecting an optimal subset that can successfully satisfy the requirements of the organization providing enough resources to the best subset of the project. The projects for which there are not enough resources or those requiring whole resources of the organization will collapse soon after failed to success. Therefore, the issue is in the risk of choosing a set of projects so that can make a balance in investment versus on collective benefit. Design/methodology/approach A model is presented for project selection and has been tested on the 37 available projects. This model could increase the efficiency of the whole subset of the project significantly in comparison to the other model and it was because of choosing a diverse subset of projects. Findings Provides a general framework for project selection and a diverse and balanced subset of projects to increase the efficiency of the selected subset. Also, reduces the impact of uncertainty risk on the project selection process. Research limitations/implications For the purposes of project selection, any project whose results are uncertain is a risky project because, if the project fails, it will reduce combined project value. For example, a pharmaceutical company’s R&D project is affected by the uncertain results of a specific compound. If the company invests in different compounds, a failure with one will be offset by a good result on another. Therefore, with selecting a diverse set of projects, this paper will have a different set of risks. Originality/value This paper discusses the risk of selecting or being responsible for selecting a project under uncertainty. Most of the projects in the field of project selection generally consider the risks facing the projects or existing models that do not take into account the risk.