Assessing the Effectiveness of ML Algorithms in Earthquake Damage Prediction

In this research paper, we examine various machine-learning algorithms to predict earthquake damage in an attempt to improve predictive performance and overall stability. We investigate the application of Random Forest, Decision Tree, SVM, KNN, and Naive Bayes algorithms and conduct a comparative study to determine which is the most suitable method to predict seismic damage cost. In this investigation, we employ an extensively engineered dataset featuring several relevant features and harmonized by preprocessing methods. Key performance points, such as accuracy and Area Under the Receiver Operating Characteristic Curve are considered during the evaluation of the examined machine-learning algorithms. This study aims to highlight the ideal approach to machine-learning algorithm selection to meet specific predictive requirements. The outcomes of this study can serve as a valuable reference for disaster management authorities when deciding on suitable algorithms to ensure precise and timely earthquake damage prediction, thereby enhancing disaster response strategies.