Prediction of changeability for object oriented classes and packages by mining change history

As a software system evolves, the classes are changed due to some development or maintenance activity, which is inevitable in software life cycle. These class changes can produce ripple effect or can lead to subsequent changes to other classes in the same package. In this paper, the changeability predictors for the classes and packages are proposed based on their past change pattern i.e. change history. Association learning method has been applied for discovering change-coupling pattern between the classes. In present work, the framework for the computation of the proposed changeability predictors is demonstrated and results are evaluated for java application. The results show that, association mining based machine learning technique can be useful to classify the classes as per their change-readiness. For doing changes in future, the proposed changeability predictors will be helpful for development team to predict the changeability of the classes.