Title: Software fault prediction using hybrid swarm intelligent cuckoo and bat-based k-means++ clustering technique
Authors: Shruti Aggarwal; Paramvir Singh
Addresses: Department of Computer Science, Sri Guru Granth Sahib World University, Fatehgarh Sahib, Punjab, India ' Department of CSE, National Institute of Technology, Jalandhar, Punjab, India
Abstract: K-means and its various hybrids are popularly used for software fault prediction. K-means++ is a hybrid clustering algorithm which overcomes major issue of getting stuck at local optima. Here, in this paper swarm intelligence-based hybrid techniques viz. cuckoo algorithm which improves the fitness function and bat algorithm which swarms with varying speeds are used on k-means++ algorithm to design a new hybrid clustering technique. KBat++ algorithm is a designed hybrid clustering technique with increased convergence rate which is further improvised using robust cuckoo swarm intelligent technique on this designed algorithm to generate CKBat++ algorithm which is predicted to generate optimised high-quality clusters. Experiments are performed using open-source UCI and promise datasets to implement and compare performance of designed algorithms with KBat and k-means++ algorithms. Accuracy, cluster quality check, CPU time etc. are used for performance comparisons. Results indicate that the designed technique which is used to predict and categorise software faults into faulty and non-faulty clusters to avoid errors and increase software reliability, is fairly better in performance than its counterparts.
Keywords: fault prediction; clustering; swarm intelligence; cuckoo algorithm; bat algorithm; k-means; k-means++ algorithm.
International Journal of Advanced Intelligence Paradigms, 2023 Vol.25 No.3/4, pp.341 - 359
Received: 05 Aug 2017
Accepted: 10 Mar 2018
Published online: 19 Jul 2023 *