Object-Oriented Class Stability Prediction: A Comparison Between Artificial Neural Network and Support Vector Machine

Software stability is an important factor for better software quality. Stable classes tend to reduce the software maintenance cost and effort. Therefore, achieving class stability is an important quality objective when developing software. Designers can make better decisions to improve class stability if they can predict it before the fact using some predictors. In this paper, we investigate the correlation between some available design measurements and class stability over versions and propose a stability prediction model using such available measurements. We conducted a set of experiments using artificial neural network (ANN) and support vector machine (SVM) to build different prediction models. We compared the accuracy of these prediction models. Our experiments reveal that ANN and SVM prediction models are effective in predicting object-oriented class stability.