Development of Data-mining Technique for Seismic Vulnerability Assessment

Assessment of seismic vulnerability urban infrastructure is an actual problem, since the damage caused by earthquakes quite significant. Despite complexity such tasks, today’s machine learning methods allow use “fast” for assessing vulnerability. The article proposes a methodology characteristics typical objects that affect their resistance; using classification and clustering methods. For analysis, we kmeans hkmeans methods, where Euclidean distance used as measure proximity. optimal number clusters determined Elbow method. A decision-making model on resistance object presented, also most important variables have greatest impact are identified. study shows results coincide with expert estimates, characteristic can be result data modeling algorithms.