Perbandingan Kinerja Model RNN, LSTM, dan BLSTM dalam Memprediksi Jumlah Gempa Bulanan di Indonesia

Abstract

Earthquakes are natural phenomena that frequently occur in Indonesia. To identify and predict the level of earthquake activity, effective prediction methods are needed. In this study, we employed a Recurrent Neural Network (RNN) to predict the average number of earthquakes that occur each month in Indonesia. This research utilized a large amount of historical earthquake data in Indonesia. We divided this data into training and testing sets to train and evaluate our prediction model. Additionally, we used Mean Absolute Error (MAE) and Mean Squared Error (MSE) as evaluation metrics to measure the accuracy of our model's predictions. The results showed that using Long Short-Term Memory (LSTM) units with a Bidirectional (BLSTM) configuration, which is a part of RNN, provided accurate predictions regarding the average number of earthquakes per month in Indonesia. We achieved an MAE of 0.0668 and RMSE of 0.0858, indicating a good level of accuracy in predicting the average number of earthquakes. This research contributes significantly to the understanding and prediction of earthquake activity in Indonesia. The use of deep learning techniques in RNN can provide accurate and reliable prediction outcomes for earthquake mitigation and risk reduction efforts in Indonesia.