Software Quality Assessment for Open Source Software using Logistic & Naive Bayes Classifier

Quality of a software product being designed, has a critical role in software process management. Detection and prediction of faults in a software with huge lines of code is a very tedious task. So it is very essential, as to reduce the maintanence cost and inturn increase the software reliability. Many object-oriented metrics have found to be suitable for software fault prediction. Using data mining techniques, design of prediction and classification models can be incorporated to give insight of the systems quality to the developing team to effectively tackle the quality problems. In this article, Chidamber and Kemerer Metric suite, along with classifiers which have immense classification capacity have been used in predicting software fault classification accuracy. From the obtained resulted it can be concluded that Logistic classifier is able to obtain better fault classification accuracy when compared to Naive Bayes approach.