Customized Particle Swarm Optimization Algorithm for Frequent Itemset Mining
N S Sukanya, P. Ranjit Jeba Thangaiah · 2020
In data analytics, frequent itemset mining is the key process for decision making. Frequent patterns are analyzed to increase the benefits, reduce the costs, maximize the income and to develop the overall resources. Every day there is an increase in data, which leads to a massive dataset. Mining the frequent itemset is a time and memory consuming task. Handling a massive transactional database is a challenging issue. There is a need for an efficient technique to retrieve the frequent patterns by reducing the computational time and memory utilization. In this paper, evolutionary computing technique is implemented to retrieve the frequent itemsets. Customized PSO is applied to mine the frequent patterns by maximizing the number frequent itemsets identified in the transactional database. First, the standard Particle Swarm Optimization (SPSO), second, Constriction Factor Particle Swarm Optimization (CFPSO), third, Customized PSO with Local Search (CPSO_LS) technique are applied to deal with the retrieval of frequent itemsets efficiently. Finally, the proposed techniques are investigated based on the number of frequent itemsets retrieved, and computational time. Experimental results illustrates that the customized PSO with local search outperforms the other two variants of the PSO algorithm.