ENHANCING CLUSTERING PERFORMANCE: A HYBRID GENERALIZED K-MEANS APPROACH

This study developed a hybrid Generalized K means clustering algorithm to boostclustering accuracy, robustness and computational efficiency across diverse datasets. The proposed method integrates multiple clustering techniques, including Forgy, Lloyd, MacQueen, Hartigan and Wong, Likas and Faber, improving initialization, assignment, and updating processes. Advanced distance metrics, particularly the Mahalanobis distance, are incorporated to account for variable correlations and variances, ensuring precise cluster assignments. The algorithm's effectiveness is validated using datasets from the World Bank Commodity Price Publication 2022 and the R console repository, including Edgar Anderson's Iris data set, COVID-19 mortality outcomes with hydroxychloroquine and chloroquine, and nicotine replacement therapy studies for smoking cessation. The methodology combines robust initialization strategies with iterative assignment and centroid update mechanisms, ensuring convergence to optimal clustering solutions. Performance comparisons with traditional K-means methods revealed the hybrid algorithm's superior accuracy, stability and efficiency, particularly in data sets with varying dimensions, distributions and complexities. By leveraging secondary data from reliable sources, the study ensures comprehensive analysis and generalization of findings. The studys findings have implications for improved pattern recognition, data segmentation and decision-making across domains, showcasing the algorithm's potential as a robust alternative to existing clustering techniques.