Akademik Veri Yönetim Sistemi
Proceedings of International Conference on Data, Electronics and Computing Vol.1 and Vol.2, Lale Özbakır,Esra Kahya Özyirmidokuz, Editör, Springer Nature, Kayseri, ss.1-6, 2025
Accurately predicting seismic events is crucial for enhancing safety measures and informing strategic infrastructure development. By leveraging data from past earthquakes, we can effectively allocate resources to mitigate potential disaster impacts. Given Turkey's active fault lines and geological position, the region is highly vulnerable to earthquakes. This research aims to conduct a comprehensive analysis of earthquake activities in Turkey and develop models using machine learning algorithms that can predict earthquake intensity. Earthquake data from the Kandilli Observatory and Earthquake Research Institute for the years 1915–2023 have been analyzed using statistical and visualization techniques. The models evaluated in this study include KNN Model (K-Nearest Neighbors), Decision Tree Model, SVM Model (Support Vector Machine), Neural Network Regression (Deep Neural Network), Bagging Model, and AdaBoost Model. As a result of the analysis, the (Deep) Neural Network Regression model was found to provide good results with an accuracy of approximately 95%. These efforts will contribute to a better understanding of earthquake risks in Turkey and improve preparedness for future seismic events.