Earthquake risk prioritization via two-step cluster analysis and SWARA-ELECTRE methods

Abstract

Earthquake is one of the most destructive disasters for people, both materially and morally. Some precautions to be taken before an earthquake reduce this harmful effect. Earthquake risk assessment is one of these precautions. Earthquake risk assessment, which is an inter-disciplinary topic, is a problem suitable for clustering and multi-criteria decision-making (MCDM) techniques, as it includes more than one criterion and alternative. In this study, decision model was proposed for earthquake risk prioritization of twenty-nine provinces with high earthquake risk in Turkey. In the proposed model, provinces were clustered via Two-Step Cluster Analysis. Indicators determined in the Two-Step Cluster Analysis were defined as criteria, and criteria weighting was made via SWARA method. After weighing the criteria, the ELECTRE I method was used for earthquake risk ranking of the clustered provinces. In the proposed model for earthquake risk assessment, the similarity of provinces can be defined, and the impact of indicators can be examined. For this purpose, as the innovative aspect of this paper, while evaluating the clustering success, it was proposed to examine the coefficients of variation for continuous variables. In the Two-Step Cluster Analysis, clusters were formed in different ways and the risk rankings for provinces divided into six sub-clusters in total were presented. As a result of the Two-Step Cluster Analysis, two clusters consisting of six provinc-es, two clusters consisting of five provinces, one cluster consisting of four provinces and one cluster consisting of three provinces were obtained. The rankings of these provinces within clusters were obtained via ELECTRE I method. The aim of the study is to guide the decision makers working on earthquake risk assessment in the practical world by providing the hybrid-ization of the specified clustering and multi-criteria decision-making methods.