Prediction of earthquakes: a step towards predicting the unpredictable

Abstract

One of the most catastrophic natural disasters is an earthquake, especially because they typically occur without warning. It has catastrophic effects on both the econ- omy of a nation and human life. Our paper is a step towards taking the challenge and complexity of predicting earthquakes and implies that the research here aims to make progress in this field. In this paper, we have proposed a hybrid model com- bining multiple algorithms which analyses already existing datasets. We have done our work in two steps. Firstly, we trained our model on multiple algorithms such as KNN, SVM, XGBOOST and ADABOOST. Then we created a hybrid model out of these algorithms which gave us the best results in terms of accuracy and precision. The use of machine learning techniques for earthquake prediction is examined in this thesis. We concentrate on the use of multiple algorithms: k-Nearest Neighbors (KNN), AdaBoost, and Support Vector Machines (SVM). We start by going over the state of earthquake forecasting right now and how machine learning is applied in this area. We next go over our study, which involves feeding our models input fea- tures made up of both seismological and geodetic data. We assess each algorithm’s performance using a range of evaluation indicators and contrast the outcomes with conventional statistical techniques. We explore the significance of our results for future research in this field and show how these machine learning techniques have the potential to be used to forecast earthquakes. Overall, this thesis makes a posi- tive contribution to the current work to increase the precision and dependability of earthquake prediction utilizing cutting-edge machine learning methods.