Correlation of change size to fault-proneness: A real life case study

Abstract

Continuous Integration (CI) is a common practice in agile software development processes. It improves the quality of the developed software and reduces the risk of integration problems. However, the effectiveness of CI depends heavily on the frequency of integration problems. For systematic improvement of the applied agile development process, the effect of large commits on the likelihood of integration problems is examined. Based on data gathered from the version control system and the CI-system of one large-scale, multi-site, agile-organized software development project in telecommunication industry, the size of commits is measured by three metrics: Count of Changed Lines, Count of Changed Files and the Count of Affected Components. As it can be found by this real life case study, there is a strong correlation between the size of a change and the likelihood of integration problems. In addition, the number of changed lines is only indirect correlated to the probability of integration problems.