Self-Organized Genetic Algorithm for Enhanced Data Clustering

This study introduces a self-organized genetic algorithm (GA) approach to address the challenge of optimizing clustering solutions in datasets with complex structures. The proposed method employs a unique population initialization strategy, wherein individuals represent clustering solutions with gender and age attributes. The iterative process involves mating, crossover, mutation, and selection to evolve and discover the optimal clustering. Self-organization principles are integrated into the GA, enhancing its performance by distinguishing leader individuals with high fitness values and active individuals. The leader individuals act as cluster heads, reducing intra-cluster distances and increasing inter-cluster distances. Empirical results demonstrate that the self-organized approach outperforms alternative techniques, particularly in terms of the average Davies-Bouldin index, showcasing its effectiveness in clustering tasks.