Title: Mining association rules using hybrid genetic algorithm and particle swarm optimisation algorithm

Authors: K. Indira; S. Kanmani

Addresses: Department of CSE, Pondicherry Engineering College, Pillaichavady, Puducherry – 605 014, India ' Department of IT, Pondicherry Engineering College, Pillaichavady, Puducherry – 605 014, India

Abstract: Evolutionary computation has become the popular choice for solving complex problems, which are otherwise difficult to solve by traditional methods. Genetic algorithm (GA) and particle swarm optimisation (PSO) are both population-based heuristic search methods, which are well suited for mining association rules. GA and PSO both have their unique features and limitations. A hybrid method combining both genetic algorithm and particle swarm optimisation called hybrid GA/PSO (GPSO) is proposed in this paper. This method is used to bring out the balance between exploration and exploitation, which will result in accurate prediction of the mined association rules and consistency in performance. GA reduces the exploitation tasks and exploration is taken care by PSO. The GPSO methodology for mining association rules performs better than the individual performance of both GA and PSO in terms of predictive accuracy and consistency when tested on five benchmark datasets in the University of California Irvine (UCI).

Keywords: genetic algorithms; particle swarm optimisation; PSO; association rules mining; predictive accuracy.

DOI: 10.1504/IJDATS.2015.067701

International Journal of Data Analysis Techniques and Strategies, 2015 Vol.7 No.1, pp.59 - 76

Available online: 18 Feb 2015 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article