Mining software code repositories and bug databases using survival analysis models

Abstract

Code repositories and bug databases contain valuable information about the process of software development. Typical studies correlate code properties with the number of faults in a software module to find error-prone modules. However, many studies do not regard the occurrence of faults over time, although the time information can be retrieved from bug databases. In order to overcome this problem, we suggest the application of survival analysis models, which are used in biostatistics and can handle time-dependent data. Because a large amount of raw data has to be evaluated statistically, we further discuss the automated retrieval and pre-processing of raw data from code repositories and bug databases.