An Exhaustive Multi-Aspect Analysis of Swarm Intelligence Algorithms in Numerical Association Rule Mining

Numerical association rule mining (NARM) is an extended version of association rule mining that determines association rules in numerical data items through distribution, discretization, and optimization methods. In the optimization techniques domain, numerous evolutionary and swarm intelligence-based algorithms have been proposed to extract association rules from numerical datasets. However, there is still a lack of comprehensive understanding regarding the performance of swarm intelligence-based algorithms, particularly for NARM. Presently, in state-of-the-art, various swarm intelligence-based optimization algorithms are claimed to be better based on their arbitrary comparisons with different algorithms in different classes, e.g., swarm intelligence-based algorithms are compared with genetic algorithms. Consequently, it becomes challenging to select the most suitable swarm intelligence-based algorithm for NARM. This article specifically aims to address this gap by conducting an exhaustive multi-aspect analysis of four popular swarm intelligence-based optimization algorithms (MOPAR, MOCANAR, ACO-R, and MOB-ARM) using four real-world datasets and six key metrics: performance time, the number of rules, support, confidence, comprehensibility, and interestingness, aiming to demonstrate the efficiencies of the SI-based algorithms in addressing the NARM problem. The achieved outcomes are also compared with the Apriori algorithm, which is one of the classical algorithms for association rule mining. In our analysis, MOPAR shows low rule count with high confidence, comprehensibility, and interestingness. MOCANAR consistently performs well across all parameters and datasets. ACO-R generates high-quality rules but may need parameter adjustments for large datasets. MOB-ARM is slower compared to others, and Apriori underperforms in support and time but excels in confidence.