<![CDATA[This study applies the K-Means and BIRCH algorithms to cluster earthquake data in Indonesia based on geographic coordinates (latitude and longitude), depth, and magnitude from 2008 to 2023. Due to its position at the intersection of three major tectonic plates, Indonesia is highly prone to earthquakes, making the mapping of vulnerable regions essential for disaster risk reduction. K-Means is selected for its simplicity and clustering effectiveness, while BIRCH is known for its scalability and efficiency in processing large datasets. The clustering process involves data preprocessing and normalization, followed by determining the optimal number of clusters using the Elbow method. Initial findings indicate that K-Means produces more distinct and well-separated clusters than BIRCH, with Silhouette Scores of 0.3501 and 0.2247, respectively. However, after expanding the dataset to 121,123 records and incorporating additional attributes such as mag_type, phasecount, and azimuth_gap, BIRCH demonstrated a significant improvement in performance, achieving a Silhouette Score of 0.3489—surpassing K-Means, which dropped to 0.1293. These results suggest that BIRCH is more effective for clustering large and complex datasets. The final clustering results are visualized on a web-based map to support spatial analysis and the identification of earthquake-prone zones.]]>
Copyrights © 2025
