A Practical Guide to Support Change-proneness Prediction

During the development and maintenance of a system of software, changes can occur due to new features, bug fix, code refactoring or technological advancements. In this context, software change prediction can be very useful in guiding the maintenance team to identify change-prone classes in early phases of software development to improve their quality and make them more flexible for future changes. A myriad of related works use machine learning techniques to lead with this problem based on different kinds of metrics. However, inadequate description of data source or modeling process makes research results reported in many works hard to interpret or reproduce. In this paper, we firstly propose a practical guideline to support change-proneness prediction for optimal use of predictive models. Then, we apply the proposed guideline over a case study using a large imbalanced data set extracted from a wide commercial software. Moreover, we analyze some papers which deal with change-proneness prediction and discuss them about missing points.