Affiliations: [a]
Department of Computer Science and Engineering, Karpagam College of Engineering, Coimbatore, India
| [b]
Department of Computer Science and Engineering, Bannari Amman Institute of Technology, Sathyamangalam, India
Correspondence:
[*]
Corresponding author. Kannimuthu Subramanian, Department of Computer Science and Engineering, Karpagam College of Engineering, Coimbatore, India. Tel.: +91 9994090779; E-mail: [email protected].
Abstract: Mining high utility itemsets (HUIs) from transaction databases is one of the current research areas in the data mining field. HUI mining finds itemsets whose utility meets a predefined threshold. It enables users to quantify the usefulness or preferences of products by utilizing different values. Since utility mining approaches do not satisfy the downward closure property, the cost of candidate generation for HUI mining in terms of time and memory space is excessive. This paper presents Genetic Algorithm based Particle Swarm Optimization (GA-PSO), which can efficiently prune down the number of candidates and optimally acquire the complete set of high utility itemsets. The proposed algorithm’s performance is assessed using the synthetic dataset T20.I6.D100K and the real-time supermarket dataset, which comprises 38765 transactions and 167 unique products. It performs very effectively in terms of time and memory on large databases constituted of small transactions, which are challenging for existing high utility itemsets mining algorithms to manage. Experiments on real-world applications show the importance of high utility itemsets in business decisions, as well as the distinction between frequent and high utility itemsets.
Keywords: Data mining, high utility itemset, genetic algorithm, particle swarm optimization, stagnation
