How Process Enactment Data Affects Product Defectiveness Prediction - A Case Study

The quality of a software product is highly influenced by the software process used to develop it. However, abstract and dynamic nature of the software process makes its measurement difficult, and this difficulty has supported the assessment insight of indirectly measuring the performance of software process by using the characteristics of the developed product. In fact, enactment of the software process might have a significant effect on product characteristics and data, and therefore, on the use of measurement and analysis results. In this article, we report a case study that aimed to investigate the effect of process enactment data on product defectiveness in a small software organization. We carried out the study by defining and following a methodology that included the application of Goal-Question-Metric (GQM) approach to direct analysis, the utilization of a questionnaire to assess usability of metrics, and the application of machine learning methods to predict product defectiveness. The results of the case study showed that the accuracy of predictions varied according to the machine learning method used, but in the overall, about 3% accuracy improvement was achieved by including process enactment data in the analysis.