Answers are grounded in this paper and cite their sources.
Estimating structural damage after an earthquake remains crucial in preventing loss of lives and properties. Conventional methods of damage estimation require a large amount of time and financial resources. For this reason, in recent years, machine learning algorithms that produce faster and more economical results have become the research topics of interest in damage estimation. Within the scope of this study, machine learning models that predict the damage level of the structure after the earthquake have been developed. In the models created, data sets containing structural and demographic information collected in 11 regions after the 2015 Gorkha, Nepal earthquake were used. Three classes of repair levels labeled by the engineers were chosen as the estimate label. The models were divided into Random Forest and XGBoost according to the classification algorithm they used, and models with and without demographic features according to the data they used. When the general accuracy rates of the models were compared, the models containing demographic information were more successful. The most successful result is the random forest model with an accuracy rate of 70.83% and the highest damage class recall value of 76.36%.
No takes yet. Share an insight, caveat, or question.
Haşıloğlu et al. (2022) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: