Scholarlink

Earthquakes are caused by rapid and sudden vibrations of the Earth’s surface, resulting from the release of energy stored in the Earth’s crust due to the sudden movement of tectonic plates or volcanic activity. therefore, four machine learning models were used to predict earthquake data: Multivariable Adaptive Regression (MARS), Nearest Neighbor Algorithm (KNN), Kernel Regression (KR), and Related Vector Machine Regression (RVM). These models were applied to earthquake data to study the effect of earthquake depth, distance, longitude, and latitude (which represent explanatory variables of earthquake strength) on earthquake strength, which is the dependent variables. The four models were compared in terms of performance using several statistical measures, including mean squared error (MSE), root mean squared error (RMSE), mean error (MAE), and coefficient of determination (R2). The results showed that the MARS model performed best in terms of error reduction and predictive accuracy, followed by the KNN model, while the KR and RVM models performed relatively poorly.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Earthquake Prediction Using Machine Learning Models (MARS, KNN, KR, RVM)

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Dumooa Mohamed Radi,
  • Sahera Hussein Zain A. L. -Thalabi

摘要

Earthquakes are caused by rapid and sudden vibrations of the Earth’s surface, resulting from the release of energy stored in the Earth’s crust due to the sudden movement of tectonic plates or volcanic activity. therefore, four machine learning models were used to predict earthquake data: Multivariable Adaptive Regression (MARS), Nearest Neighbor Algorithm (KNN), Kernel Regression (KR), and Related Vector Machine Regression (RVM). These models were applied to earthquake data to study the effect of earthquake depth, distance, longitude, and latitude (which represent explanatory variables of earthquake strength) on earthquake strength, which is the dependent variables. The four models were compared in terms of performance using several statistical measures, including mean squared error (MSE), root mean squared error (RMSE), mean error (MAE), and coefficient of determination (R2). The results showed that the MARS model performed best in terms of error reduction and predictive accuracy, followed by the KNN model, while the KR and RVM models performed relatively poorly.