Predictive Modelling of Seismic Hazard Applying Naïve Bayes and Granular Computing Classifiers

Seismic damage mapping is an imperative part of urban risk assessment and reduction plans, especially in Tehran, the capital of Iran, where a large population lives in a potentially active seismic area. In this paper, a Bayesian statistical classification has been compared with the granular computing (GrC) algorithm for seismic physical vulnerability assessment. Both classifiers are verified by accuracy measurements. The results show that GrC had a better performance for seismic vulnerability assessment. In addition, the city of Tehran is judged to have a severe situation against possible earthquakes.