Machine learning methods for predicting earthquake frequency in the geographical region surrounding northern Morocco

¹ Laboratory of Applied Sciences (LSA), National School of Applied Sciences, Abdelmalek ESSAADI University, P.O Box 03, Ajdir Al-Houceïma, Morocco.
² Department of Computer Science, Faculty of Sciences and Techniques, Errachidia, Morocco.
³ Laboratory of Solid-State Physics, Department of Physics, Faculty of Science Dhar El Mahraz, Sidi Mohamed Ben Abdellah University, Fez, Morocco.

Published online: 13 May 2026

Abstract

In order to reduce risks, we utilize machine learning/deep learning models to forecast the frequency of earthquakes in a specific geographic region, such as northern Morocco. Several types of studies have allowed the obtaining of acceptable results by integrating the ARIMA model with machine learning/deep learning models, such as LSTM, XGBoost, SVR and RF. Given that Morocco is situated in a moderately active seismic zone, our study examines how well machine learning and deep learning models predict the frequency of earthquakes in the area surrounding northern Morocco. Knowing that Morocco is located in a moderately active seismic zone, our study compares the effectiveness of machine learning and deep learning models in predicting the frequency of earthquakes in the region around northern Morocco. We employ a collection of hybrid models that integrate the ARIMA models with various machine learning/deep learning models to operate as a guiding core for future development challenges, particularly because it has allowed us to perform a large degree of prediction. We obtained very significant results regarding hybrid ARIMA models and machine learning models (RF, SVR, XGB), whilst the ARIMA model failed when used on its own or even when hybridised with the deep learning model LSTM.