Exploring the relationships between design measures and change proneness in object-oriented systems

Abstract

Managing change early in software development life cycle can lead to good quality software. Developing software requires a huge amount of time and assets. On the other hand, maintenance expenses are higher than development expenses, which are a major cause for concern, especially, for the systems developed with new technologies. Predicting change is an active research area in the field of software engineering. Such predictions can be used to estimate changes from one release of software to the next. This paper explores various parameters and relationships that can be considered in predicting change prone classes. We elaborate dependency approach method which calculates if one class is dependent on other or not, if a class is dependent, dependency value can be calculated using UML 2.0 sequence diagram. Further, we have extracted frequency and popularity feature from an application that can be used for estimating change prone classes. We have related frequency and popularity with dependency in terms of method calling. Also, we have created a tool in Java which generates UML 2.0 sequence diagram and using the sequence diagram dependency value is generated.