This study aims to enhance the K-Means clustering algorithm to improve the analysis of earthquake occurrence patterns in the Philippines. Traditional K-Means, while effective, suffers from limitations such as random initialization and slow convergence. To address these issues, we propose an improved K-Means algorithm that strategically selects initial centroids based on a distance-weighted probability distribution to enhance accuracy and processes data in smaller batches to reduce computation time, thereby improving scalability and convergence speed. Using earthquake data from the Philippine Institute of Volcanology and Seismology (PHIVOLCS), we evaluate the performance of the enhanced algorithm using metrics such as Silhouette Score and Time Complexity. Results demonstrate that the proposed modifications significantly enhance clustering accuracy, computational efficiency, and scalability, leading to more precise identification of high-risk seismic areas. By providing a more accurate and efficient framework for seismic data analysis, this research contributes to disaster preparedness, risk mitigation, and informed decision-making in urban planning and disaster management.