A comprehensive analysis and prediction of earthquake magnitude based on position and depth parameters using machine and deep learning models

Earthquake is one of the major natural disasters that not only costs human lives but also leads to financial losses, which affect the country’s economy. Earthquake Prediction is one of the challenging research areas because its early prediction can save a lot of human lives, helps in minimizing the financial losses to some extent. The objective of this research is to develop an earthquake prediction model based on the position and depth by using machine learning and deep learning algorithms. The dataset is split into seven different csv files after thorough processing and a requisition of best-performing regression models is done to compute the results. These algorithms include Random forest (RF) Regression, Multi-Layer Perceptron (MLP) regression, and Support Vector Regression (SVR). The method is applied for different radii around the target. The dataset for this research is taken from the USGS website. The efficiency of algorithms is compared by computing the deviation between actual and predicted outcomes by using the error metrics. The results are evaluated using the Root Mean Square Error (RMSE) metric. Considering the boundary values, the RMSE for RF Regression is 1.731, for MLP regression the value is 1.647 and for SVR the RMSE achieved is 1.720, all for a minimum radius value of 100 and similarly 0.436, 0.428 and 0.449 RMSE is achieved for the respective algorithms on a maximum radius of 5000. The results demonstrate that MLP Regressor is performing better than other algorithms as the error is least in the case of this algorithm.