Forecasting Earthquake Using Machine Learning

Abstract

Many lives and properties were lost in the past due to unforeseeable deadly earthquakes in the Philippines, which encouraged the researcher to examine different models to achieve the best model utilizing machine learning to forecast earthquakes. The researcher employed ARIMA as the baseline model for DNN, RNN, LSTM CNN, and CNN+LSTM then compared neural networks to determine which model had the lowest error using MEA - mean absolute error. After comparing the MEA from the various models, LSTM had the lowest mean absolute error, implying that it is the best model for forecasting earthquakes.