GS-FFO Based Meta-Heuristic Clustering for Discovering Colossal Patterns in High-Dimensional Data

Colossal pattern mining, which involves deriving important relationships and insights from large and complex datasets, is an essential part of data analysis. Extracting association rules that satisfy preset minimum support and confidence criteria from a given database is the aim of association rule mining (ARM). Nevertheless, ARM often generates an unusually large number of association rules that are hard for end users to comprehend or validate, which lessens the significance of data mining discoveries. Based on the clustering methodology, this research study presents a new mining strategy. The process begins with a method called the Expressed Data Matrix (EDM), which transforms unstructured data into a structured format and enhances the quality and organisation of the information. Primary data cleaning processes are used to eliminate noise and inconsistencies from the data in order to further improve it. To address high-dimensionality issues and boost mining productivity, dimensionality reduction—which lowers the dataset to its most valuable features—is accomplished using Eigen-Principal Component Analysis (PCA). The core of this study is the hybridised clustering approach, or GS-FFO, which combines the Fruit Fly Optimisation technique (FOA) with Glow Swarm Optimisation (GSO). This technique improves the quality of clustering and reduces the probability of local optima convergence by integrating the benefits of both approaches. The suggested GS-FFO model outperforms with four datasets, including accident, pumbs*, retail, and yeast, with an accuracy of up to 98.5%.