Software Defect Prediction using Outlier Detection Algorithm

In the dynamic landscape of software development, the pursuit of creating flawless and efficient software is a perpetual challenge. However, the presence of software defects stands as an inherent reality in this complex process. Software Defect Prediction (SDP) has been a common topic in study literature. Different methods and approaches have been developed to do the prediction in different datasets. In this research, outlier detection algorithms are used to conduct experiments in SDP. Isolation Forest, Local Outlier Factor, and One-Class SVM are the three outlier detection models used in this research. The experimental results show that One-Class SVM has better performance than the other models with the F1 score reaching 0.9101 and accuracy up to 0.84.