Improving change proneness prediction in UML based design models using ABC algorithm

In the field of software engineering, which is emerging as the undisputed man of the match in the ever-changing sports of sophistications, the incessant changes effected in software every day have assumed such an alarming proportion causing untold and unimagined paradoxes that it is highly essential to initiate instant and immediate steps to balance this blitz. It does not mean that no endeavor has been made in the bye-gone era to tackle the issue. In fact, several methods to solve this dilemma were introduced in the past by predicting the changes in the software. But these methods have miserably met with a waterloo in facilitating a highly appreciable forecast harvest. To handle this problem with an eye on good prediction accuracy, a new method is introduced in our paper. The two phases in our proposed work are: (1) Feature Identification (2) Classification of classes for the change-proneness prediction. In feature identification phase, the features obtained from any input application are time, trace events, behavioral dependency, frequency and popularity, which help to predict the change proneness in our work. There are three ways by which these five features are found from the application. They are: Features obtained directly from an application such as time, trace events; Features obtained from UML Diagrams such as behavioral dependency and Features obtained from optimal frequent item set mining and ABC such as frequency and popularity. Thus all these five features are obtained from our proposed work and then in the classification phase, these features are given as the input to the ID3 decision tree algorithm for effectively classifying the classes according as 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.