Comparative Analysis of Centroid-Based and Density-Based Clustering for Indonesian Earthquake Data

Jalaludin Shofa Mufti, Arian Dhini

Research output: Contribution to journalConference articlepeer-review

Abstract

Clustering characteristics of earthquakes in Indonesia from 2004 to 2023 are analyzed using data from the United States Geological Survey (USGS). Earthquake clusters are identified using two distinct algorithms: K-Means, a centroid-based approach, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN), a density-based method. Multiple cluster characteristics are compared, focusing on magnitude, depth, and location represented by latitude and longitude. The analysis suggests that both clustering methods effectively identify patterns in Indonesian seismic data, with K-Means slightly outperforming DBSCAN based on silhouette coefficient and Davies-Bouldin index evaluations. Results show distinct regional patterns in earthquake characteristics, particularly in terms of depth, across the Indonesian archipelago. K-Means identifies three clusters, while DBSCAN produces two clusters and a noise category, with the noise category in DBSCAN sharing similarities with the deep-earthquake cluster identified by K-Means. The observed clustering patterns provide important insights into the seismic behavior of different regions within Indonesia and can be used to improve local seismic hazard assessments and disaster management strategies.

Original languageEnglish
Pages (from-to)464-467
Number of pages4
JournalInternational Conference on Computer, Control, Informatics and its Applications, IC3INA
Issue number2024
DOIs
Publication statusPublished - 2024
Event11th International Conference on Computer, Control, Informatics and its Applications, IC3INA 2024 - Hybrid, Bandung, Indonesia
Duration: 9 Oct 202410 Oct 2024

Keywords

  • Cluster Analysis
  • DBSCAN
  • Earthquake
  • K-Means
  • Seismicity

Fingerprint

Dive into the research topics of 'Comparative Analysis of Centroid-Based and Density-Based Clustering for Indonesian Earthquake Data'. Together they form a unique fingerprint.

Cite this