MSCPDPLab: A MATLAB toolbox for transfer learning based multi-source cross-project defect prediction

Abstract

Software defect prediction (SDP) plays an important role in allocating testing resources and improving testing efficiency. Multi-source cross-project defect prediction (MSCPDP) based on transfer learning refers to transferring defect knowledge from multiple source projects to the target project. MSCPDP has drawn increasing attention from academic and industry communities, and some MSCPDP methods have been proposed. However, most existing MSCPDP models are not open-source. MSCPDPLab replicates nine state-of-the-art MSCPDP models with unified interface and integrates the processes of data loading, model training and testing, and performance evaluation (including 13 performance measures). This paper describes the toolbox’s functionalities and presents its ease of use.

Keywords

• Multiple source datasets
• Cross-project defect prediction
• Mining software repository
• Transfer learning