An Approach for Earthquake Magnitude Prediction in Sumatra using Linear Regression and Random Forest Algorithms

Abstract

Sumatra is one of the largest islands in Indonesia and is highly susceptible to seismic activity due to its location between the Indo-Australian and Eurasian tectonic plates. Many significant earthquakes have occurred in Sumatra, resulting in substantial losses and damage. Despite advances in seismic monitoring technology and methods, achieving high accuracy in earthquake prediction remains a major challenge. This study presents a machine learning approach to predicting earthquake magnitudes in Sumatra using Random Forest and Linear Regression algorithms. Utilizing a comprehensive dataset from the United States Geological Survey (USGS) covering seismic events in Sumatra from July 1960 to April 2024, this research aims to improve the accuracy of magnitude predictions. This dataset includes variables such as time, location, depth, latitude, longitude, and magnitude. The study employs Python libraries for data visualization and correlation analysis. Visualizations show a strong correlation between predicted and actual earthquake magnitudes, mapping seismic events along well-known fault lines such as the Sunda Megathrust and Sumatran Fault. The results of this study demonstrate the potential of machine learning models to enhance seismic hazard assessment strategies, significantly contributing to more effective disaster management and mitigation practices in earthquake-prone areas.