Enhanced Grid Based K-Medoids Clustering Technique for Customer Segmentation

Among the existing clustering algorithms, the K-Medoids algorithm is widely used due to its robustness and effectiveness. However, its performance can be affected by the selection of initial cluster centers and sensitivity to noise. To address these issues, this paper proposes an improved K-Medoids clustering algorithm. A density-based noise removal technique is introduced through Grid to filter out outliers and obtain regional density. Grid-based clustering with K-Medoids improves efficiency by reducing the number of distance calculations through space partitioning. It is robust to noise and outliers since KMedoids selects actual data points as representatives. The approach reduces computational complexity by clustering grid cells instead of all data points. Additionally, the influence of density estimation errors on centre selection is reduced through granularity-based adjustments, minimizing manual intervention in the peak density algorithm. This approach enhances clustering accuracy and stability