Early Earthquake Prediction Using Machine Learning Technique for North Part of India Saving Human Life

Abstract

An earthquake is a powerful natural disaster that causes the earth's surface to tremble abruptly. Earthquakes harm infrastructure and buildings, which impacts daily life. The originality of this work is that machine learning can be a powerful tool for early seismic impact prediction. With the use of six different machine learning classifiers—Artificial Neural Network, Random Tree, CHAID, Discriminant, XGBoost Tree, and Tree-AS—and six datasets from different regions of India, this understanding was evaluated. Every method has been applied to every dataset.

The study intends to forecast future earthquake magnitudes in India and the surrounding areas using historical earthquake data. With 98.20% accuracy for the Himachal Pradesh dataset, 89.10% for the UttraKhand dataset, 99.13% for the North India dataset, 98.21% for the North East India dataset, 97.50% for the Uttar Pradesh dataset, and 90.00% for the Nearby India's Country dataset, XGBoost Tree was shown to have the highest accuracy. All things considered, the XGBoost tree classifier did well across the majority of datasets.