Exploring the relationship of a file's history and its fault-proneness: An empirical study

Knowing which particular characteristics of software are indicators for defects is very valuable for testers in order to allocate testing resources appropriately. In this paper, we present the results of an empirical study exploring the relationship between history characteristics of files and their. defect count. We analyzed nine open source java projects across different versions in order to answer the following questions: 1) Do past defects correlate with a file's current defect count? 2) Do late changes correlate with a file's defect count? 3) Is the file's age a good indicator for its defect count? The results are partly surprising. Only 4 of 9 programs show moderate correlation between a file's defects in previous and in current releases in more than the half of analysed releases. In contrast to our expectations, the oldest files represent the most fault prone files. Additionally, late changes influence file's defect count only partly.