Predicting Earthquake Damage and Rehabilitation Intervention Using Adaptive Fuzzy C-Means-Based Support Vector Machine

Earthquakes are the greatest rates of human loss among natural disasters in a past 20 years. Predicting rehabilitation intervention and damage grades is significant, especially in the moderate aftermath of a strong earthquake as prioritized in post-earthquake housing rescue requires data about damage extent. However, building collapses caused by earthquakes lead to huge loss of property and life. Therefore, an Adaptive Fuzzy C-means-based Support Vector Machine (AFCM-SVM) is proposed to predict earthquake damage and rehabilitation intervention using Machine Learning (ML). Initially, the xBD dataset is employed to evaluate the proposed technique and min-max normalization is established which enhances numerical stability. The AFCM is utilized to segment the building effectively and Gray-Level Co-occurrence Matrix (GLCM) is employed for extraction. Finally, SVM is performed to predict the earthquake damage and rehabilitation intervention. When compared with existing approaches like Convolutional Neural Networks (CNN) and Siamese Neural Networks, the AFCM-SVM achieves a better f1-score of 0.945 respectively.