Major earthquake event prediction using various machine learning algorithms

Roxane Mallouhy (1) , Chady Abou Jaoude (1) , Christophe Guyeux (2) , Abdallah Makhoul (2)

1 PMU - Prince Mohammad Bin Fahd University
2 FEMTO-ST - Franche-Comté Électronique Mécanique, Thermique et Optique - Sciences et Technologies (UMR 6174)

At least two basic categories of earthquake prediction exist: short-term predictions and forecast ones. Short term earthquake predictions are made hours or days in advance, while forecasts are predicted months to years in advance. The majority of studies are done on forecast, taking into consideration the history of earthquakes in specific countries and areas. In this context, the core idea of this work is to predict whereas an event is classified as negative or positive major earthquake by applying different machine learning algorithms. Eight different algorithms have been applied on a real earthquake dataset, namely: Random Forest, Naive Bayes, Logistic Regression, MultiLayer Perceptron, AdaBoost, K-nearest neighbors, Support Vector Machine, and Classification and Regression Trees. For each selected model, various hyperparameters have been selected, and obtained prediction results have been fairly compared using various metrics, leading to a reliable prediction of major events for 3 of them.

Domaines: Calcul parallèle, distribué et partagé [cs.DC]; Cryptographie et sécurité [cs.CR]; Technologies Émergeantes [cs.ET]; Génie logiciel [cs.SE]; Informatique ubiquitaire; Système multi-agents [cs.MA]; Modélisation et simulation

Roxane Mallouhy, Chady Abou Jaoude, Christophe Guyeux, Abdallah Makhoul. Major earthquake event prediction using various machine learning algorithms. International Conference on Information and Communication Technologies for Disaster Management, Dec 2019, Paris, France. ⟨hal-02952670⟩