Code decay analysis of legacy software through successive releases

Prediction of problematic software components is an important activity today for many organizations as they manage their legacy systems and the maintenance problems they cause. This means that there is a need for methods and models to identify troublesome components. We apply a model for classification of software components as green, yellow and red according to the number of times they required corrective maintenance over successive releases. Further, we apply a principal component and box plot analysis to investigate the causes for the code decay and try to characterize the releases. The case study includes eight releases and 130 software components. The outcome indicates a large number of healthy components as well as a small set of troublesome components requiring extensive repair repeatedly. The analysis characterizes the releases and indicates that it is the relationship between components that causes many of the problems.