Guidelines for the application of data envelopment analysis to assess evolving software

Abstract

The assessment of software evolution in terms of quality poses significant challenges as different metrics have to be combined and normalized over the size of each examined version. Data Envelopment Analysis (DEA), a non-parametric technique from production economics, can offer a unified view of several design properties providing insight into global evolutionary trends. In this paper a set of practical guidelines for the application of DEA and the interpretation of the extracted results is proposed, with a focus on open source software, where limited information and documentation might be available. © Springer-Verlag Berlin Heidelberg 2014.