Paper Title
A Compound Dagging Approach for Identifying Defective Classes in Object-Oriented Design

Abstract
Software defect prediction (SDP) has risen as an innovative research area in determining defective classes among object-oriented (OO) sub-systems at early progression stage in software life cycle. It is feasible to identify the faulty classes with certain object-oriented metrics to help the developers foresee the defects. Various classification techniques have been used for predictive analysis on software error detection. Amongst them, ensemble methods have emerged as more accurate and statistically efficient approach than conventional classifiers. In this paper, twenty object-oriented metrics were studied. Additionally, some well-known supervised machine learning algorithms along with compound dagging approach (CDA) were used on twelve object-oriented projects. The analysis shows the dominance of dagging approach over the conventional algorithms in terms of accuracy, f-score, recall and precision. Keywords - Object-oriented metrics, compound dagging approach, ensemble, software defect prediction.