Data envelopment analysis based multi-objective optimization model for evaluation and selection of software components under optimal redundancy

Abstract

Software developers face the challenge of developing in-time, low cost, high profit and high-quality software to meet competitive requirements and user demands. The software components for the same can be selected either from the available commercial-off-the-shelf repository or developed in-house. In this paper, we propose a data envelopment analysis (DEA) based nonlinear multi-objective optimization model for selecting software components in the presence of optimal redundancy to ensure software reliability. The proposed optimization model integrates both build and/or buy decisions for selection of components. We use DEA technique for evaluating the fitness of software components based upon multiple inputs and outputs provided by various members of the decision group. The overall efficiency score of each software component is obtained from the aggregated information. The proposed optimization model minimizes the total cost of software system and maximizes the total value of purchasing using constraints corresponding to compatibility of selected components, reliability, execution time, and delivery time of the software system. It also provides the information on the testing efforts needed to be performed on in-house developed components. A real-world case study of modular software development is discussed to illustrate the efficiency of the proposed optimization model. To the best of our knowledge, there exists no previous study on integrated optimization model for the software component selection problem involving build and/or buy decisions under optimal redundancy.