Comparison of DBSCAN and PCA-DBSCAN Algorithm for Grouping Earthquake Area

Abstract

Geologically, the territory of Indonesia is where the three active tectonic plates meet which are always moving and colliding with each other, resulting in earthquakes, volcanic pathways, and faults. Earthquake is a natural disaster that cannot be avoided or prevented, but the consequences of earthquakes can be minimized. Based on data obtained from Meteorology, Climatology and Geophysics Agency (MCGA), earthquakes often occur in Indonesia. Data obtained from earthquakes can be grouped to map the area of earthquake occurrence and an analysis will be carried out to determine the characteristics of earthquake clustering areas. The clustering in this is study conducted with two experiments, first experiment is Density-Based Spatial Clustering of Applications with Noise (DBSCAN) without dimensional reduction and second experiment is DBSCAN clustering with dimensional reduction using Principal Component Analysis (PCA). The best cluster results can be found by calculating the value of Silhouette Index (SI) of each cluster. From the two experiments, the highest SI value was obtained in experiment using PCA, which was 0.4137. Then the second experiment was used as the best cluster results with the highest Dept and Magnitude features in clusters 19 and 17 which showed the 5 main regions where earthquakes often occur are Sumatra, Banda Sea, Moluccan Sea, Irian Jaya and Sulawesi