A Metaheuristic Perspective on Extracting Numeric Association Rules: Current Works, Applications, and Recommendations

In the vast field of data mining, the increasing significance of Numerical Association Rule Mining (NARM) lies in its capacity to unearth recurrent patterns and correlations across diverse attribute types, resonating across multifarious sectors such as healthcare, commercial databases, and beyond. This article explores in depth the intricacies of optimization algorithms and metaheuristic approaches within the NARM framework, highlighting their essential role in amplifying the effectiveness and computational efficiency of the algorithms developed. In particular, the integration of metaheuristic optimization appears to be a significant advance, improving the accuracy and reliability of derived rules while avoiding the computational rigors of conventional processes. Exploration in this study, covers various areas of association rules, including numerical, fuzzy and high-utility sets, providing a solid synthesis of a meta-study and offering a holistic view that interweaves historical, methodological and future-oriented perspectives, thus seeking to immerse future research efforts in a comprehensive understanding of the incessant optimization approaches inherent in NARM’s vast scope in data mining. In particular, this survey considered the extensive metaheuristic-based NARM research works between 2015 and 2023. Initially commencing with a large corpus of 19,500 papers, a stringent filtration process was employed, resulting in the identification of 180 pertinent papers that contributed significantly to this survey.