Analysis of Seismic Data Using Partition-Based Clustering Techniques
Rohan Alom, Anirban Mazumdar, Rabinder Kumar Prasad, George Basumatary, Bidisha Baruah · 2022 IEEE Global Conference on Computing, Power and Communication Technologies (GlobConPT) · 2022
Cluster analysis is a machine learning technique that identifies patterns in a given dataset. This study uses partition-based clustering techniques to perform unsupervised seismic event classification based on spatial location, magnitude and depth. A global earthquake catalogue dataset from 1970 to 2014 is being used. We investigate five partition-based clustering techniques, namely k-means, Partition Around Medoids (PAM), k-means++, Clustering Large Applications (CLARA) and Fuzzy c-means, using three widely used validation indices namely Sum of Squared Errors (SSE), Davies-Bouldin Index (DBI), and Dunn Index (DI). The attributes considered in this context are longitude, latitude, magnitude, and depth. So, in the experimental analysis based on validation indices, k-means has outperformed its competitors. Similarly, k-means++ comes second among its competitors.