Clustering algorithms are critical in data analysis and optimization, particularly for segmenting datasets in practical applications. This study proposes an improved variant of K-means clustering algorithm and do evaluation on the performance of nine popular clustering methods—DBSCAN, HDBSCAN, Spectral Clustering, Hierarchical Clustering, OPTICS, Mean-Shift Clustering, Self-Organizing Map (SOM), K-Means, Gaussian Mixture Model (GMM), and the proposed Improved K-Means algorithm—using Vehicle Routing Problem (VRP) dataset. Implemented in Python, the methods are compared using the Calinski-Harabasz Index and Silhouette Index to assess cluster quality. Results show that the proposed improved K-Means algorithm, with its optimized parameter configuration through the initialization process of the parameter K and cluster centroid set, performs exceptionally well compared to other methods. These findings highlight the importance of parameter tuning and algorithm selection in clustering tasks, offering valuable insights for researchers and practitioners. This work provides practical guidance for selecting and refining clustering approaches, contributing to advancements in applications such as vehicle routing and beyond.