Enhancing Frequency Based Change Proneness Prediction Method Using Artificial Bee Colony Algorithm

In the field of software engineering, during the development ofObject Oriented (OO) software, the knowledge of the classes which are more prone tochanges in software is an important problem that arises nowadays. In order to solve this problem, several methods were introduced by predicting the changes in the software earlier. But those methods are not facilitating very good prediction result. This research work proposes a novel approach for predicting changes in software. Our proposed probabilistic approach uses the behavioral dependency generated from UML diagrams, as well as other code metrics such as time and trace events generated from source code. These measures combined with frequency of method calls and popularity can be used in automated manner to predict a change prone class. Thus all these five features (time, trace events, behavioral dependency, frequency and popularity) are obtained from our proposed work. Then, these features are given as the input to the ID3 (Interactive Dichotomizer version 3) decision tree algorithm for effectively classifying the classes, whether it predicts the change proneness or not. If a class is classified into prediction of change prone class, then the value of change proneness is also obtained by our work.