Time evolution and distribution analysis of software bugs from a complex network perspective

Successful software systems are constantly under development. Since they have to be updated when new features are introduced, bug are fixed and the system is kept up to date, they require a continuous maintenance. Among these activities the bug fixing is one of themost relevant, because it is determinant for software quality. Unfortunately, software houses have limited time and developers to address all these issues before the product delivery. For this reason, an efficient allocation of these resources is required to obtain the quality required by the market. The keyword for a correct management of software product process is measure. As De-Marco states “you cannot control what you cannot measure”, and this thesis is mainly devoted to this aspect. This dissertation bears with software measures related to bug proneness and distribution analysis of software bugs. The aim is to describe the bug occurrence phenomena, identify useful metrics related to software bugs proneness and finally to characterize how bug population is distributed and evolve, discussing also the model able to explain this evolution. Studying the relationship between code evolution and bug distribution or bug-proneness, we foresee which software structure will come out. Thus, this research provides information and guidelines tomanagers, helping them to plan, schedule activities and allocate resources, during software development.