Using data envelopment analysis for evaluating alternative software development process configurations

Summary form only, as given. The goal of achieving better software depends on improvements in software development processes. This paper provides a nonparametric, quantitative methodology for evaluating alternative software development process configurations as a tool for process improvement. The methodology uses data envelopment analysis (DEA) for comparing the impact of varying degrees of software inspection on project schedule, cost and quality. Since every phase of the development process can be inspected at varying levels, a large number of process combinations become possible. Thus a manager´s job to compare and select an efficient process configuration can become time consuming. This is especially the case when several process configurations are being evaluated by comparing several productivity measures. Fortunately, data envelopment analysis (DEA) can be applied as a method to reduce the long list of candidates for best practices to a selected few process options for further analysis. In this study, 257 different process options were compared against each other. The input and output values for each process option were estimated and compared using DEA. Of this quantity, 7 models were identified as candidates for best practices. An extension to this model incorporated managerial preferences in the form of weight restrictions applied to input/output variables