Title: Generalised rule induction-based model for software fault prediction
Authors: Ashutosh Mishra; Meenu Singla
Addresses: Computer Science and Engineering Department, Thapar University, Patiala, Punjab, India ' Computer Science and Engineering Department, Thapar University, Patiala, Punjab, India
Abstract: Software fault predictions have great importance during maintenance and evolution of softwares. To upgrade the quality of the software, it is necessary to predict the faulty software modules. Previous research investigated was mainly focused on the binary classification of software class modules using different fault prediction techniques. However, much less work was done on the prediction of the number of faults. In this study, a descriptive technique called as Generalised Rule Induction (GRI) associations rule mining is proposed to identify the number of faults in the faulty class module. The proposed technique is implemented for five releases of open source Apache Ant project which is taken from PROMISE repository. The results show that the generated rules for the class modules containing single fault achieves better accuracy with fewer rules.
Keywords: software fault prediction; dependent variable; independent variable; clustering; association rule mining.
International Journal of Reliability and Safety, 2021 Vol.15 No.1/2, pp.89 - 103
Received: 25 Mar 2021
Accepted: 25 Apr 2021
Published online: 13 Dec 2021 *