Prediction of software defects using Twin Support Vector Machine

Considering the current scenario, the crucial need for software developer is the generous enhancement in the quality of the software product we deliver to the end user. Lifecycle models, development methodologies and tools have been extensively used for the same but the prime concern remains is the software defects that hinders our desire for good quality software. A lot of research work has been done on defect reduction, defect identification and defect prediction to solve this problem. This research work focus on defect prediction, a fairly new filed to work on. Artificial intelligence and data mining are the most popular methods researchers have been using recently. This research aims to use the Twin Support Vector Machine (TSVM) for predicting the number of defects in a new version of software product. This model gives a nearly perfect efficiency which compared to other models is far better. Twin Support Vector Machine based software defects prediction model using Gaussian kernel function obtains better performance as compare to earlier proposed approaches of software defect prediction. By predicting the defects in the new version, we thereby attempt to take a step to solve the problem of maintaining the high software quality. This proposed model directly shows its impact on the testing phase of the software product by simply plummeting the overall cost and efforts put in.