Prediction of Fault-Prone Classes Using the UML Class Diagram

Complexity is an important quality attribute. Software complexity can be measured in design phase may produce good quality product.In this paper,we measure the complexity of object-oriented system at design phase to predict the fault-prone classes.The facility to predict the fault-prone classes can provide direction for software testing and improve the efficiency of development process. We built the Naïve Bayesian and k-Nearest Neighbors model to find the relationship between the design complexity and fault-proneness. The proposed models are empirically evaluated using four version of JEdit. The models are validated using 10fold cross validation. The performance of prediction models were evaluated by goodness of fit criteria and Receiver Operating Characteristic (ROC) investigation. Results obtained from our casestudy shows the average of models developed by design complexity can predict upto 70% faulty classes in object oriented system. It is a better an early indicator of software quality and make the software as reusable as well as maintainable.