Discovering determinants of high volatility software
Carolyn Budinger Seaman, A. Güneş Koru, Sreedevi Sampath · 2009
This topic paper presents a line of research that we are proposing that incorporates, in a very explicit and intentional way, human and organizational aspects in the prediction of troublesome (defect-prone or change-prone or volatile, depending on the environment) software modules. Much previous research in this area tries to identify these modules by looking at their structural characteristics, so that increased effort can be concentrated on those modules in order to reduce future maintenance costs. The outcome of this work will be a set of models that describe the relationships between software change characteristics (in particular human, organizational, and process characteristics), changes in the structural properties of software modules (e.g. complexity), and the future volatility of those modules. The impact of such models is two-fold. First, maintainers will have an improved technique for estimation of maintenance effort based on recent change characteristics. Also, the models will help identify management practices that are associated with high future volatility.