ABSTRACT
The high throughput and good quality software are developed if effective defect prone modules can be predicted by testers at early stages. This enables the testers to focus on the activities, allocation of effort, and efficient resource management. Accurate prediction of fault inclined modules in software development manner allows effective detection and identity of defects. The existing researches used support vector machine (SVM) in standalone mode in the sense that no pre-processing is performed before invoking classification; thus degrading the accuracy of prediction. This chapter delivered a singular approach combining SVM and feature choice for software fault proneness prediction (SFPP). Experiments suggest that the accuracy of the proposed method using five object-oriented (OO) metrics, that is, line of code (LOC), program level (L), branch count (BR), and unique operands. Hence, the OO metrics have a large prone in defect prone modules.
