Defect prediction using social network analysis on issue repositories

Abstract

Researchers have been building intelligent oracles to predict defect proneness of software products. These models use product and process related attributes as their input and their intelligence come from the machine learning algorithms they employ. In recent years researchers emphasize that the algorithms reached a ceiling that it is not worth the e ort in working to increase their performance and information content of input data should be enriched to include di erent kinds of metrics to eliminate this ceiling e ect. One set of metrics is people related metrics. People are the most primary elements of software development process. It is critical to understand how they interact with each other and how these interactions a ect the quality of the end product in terms of defects. In this research we propose to include a new set of metrics, a.k.a. social network metrics on issue repositories in predicting defects. This process consisted of two di erent stages. First of all we observed relation between communication network structure and number of defects xed in a time period during development of software. We were unable to nd a signi cant relation in this observation. Then we conducted le level defect prediction using social network metrics. Results of our proposed model revealed that compared to other set of metrics such as churn metrics using social network metrics on issue repositories either considerably decreases high false alarm rates without compromising the detection rates or considerably increases low prediction rates without compromising low false alarm rates. Therefore we recommend practitioners to collect social network metrics on issue repositories since people related information is a strong indicator of past patterns in a given team.