Penguins Search Optimisation Algorithm for Association Rules Mining

• Published in: Journal of computing and information technology Vol. 24; no. 2; pp. 165 - 179
• Main Author: Gheraibia, Youcef
• Format: Journal Article Paper
• Language: English
• Published: Sveuciliste U Zagrebu 01.06.2016; Fakultet elektrotehnike i računarstva Sveučilišta u Zagrebu; University of Zagreb Faculty of Electrical Engineering and Computing
• Subjects: Algorithms; Analysis; ARM; association rules mining; biological data-set; Data mining; Engineering research; Information systems; Mathematical optimization; Methods; overlap measure; penguins search optimisation algorithm
• ISSN: 1846-3908; 1330-1136; 1846-3908
• DOI: 10.20532/cit.2016.1002745

Association Rules Mining (ARM) is one of the most popular and well-known approaches for the decision-making process. All classic exhaustive ARM algorithms are time consuming and generate a very large number of association rules, even the recent proposed meta-heuristics based methods generate a small number of high quality rules but with high overlapping. To deal with this issue, we propose a new ARM approach based on penguins search optimisation algorithm (Pe-ARM for short). Moreover, an efficient measure is incorporated into the main process to evaluate the amount of overlapping among the generated rules. The proposed approach also ensures a good diversification over the whole solutions space. To demonstrate the effectiveness of the proposed approach, several experiments have been carried out on different data sets and specifically on the biological ones. The results reveal that the proposed approach outperforms the well known meta-heuristics ARM algorithms in both execution time and solution quality.