An empirical seismic vulnerability model.

This paper proposes a novel seismic vulnerability model for the classification of the existing residential building stock. The vulnerability model rests on a data-driven approach, taking advantage of observed seismic damages detected on several Italian building typologies, struck by the 2009 L'Aquila earthquake. Unsupervised machine learning techniques are exploited for clustering empirical damage data and objectively identifying vulnerability classes of decreasing vulnerability. The cascading use of different strategies, involving clustering analysis and probability theory, results in a comprehensive vulnerability model, which allows for determining, into a probabilistic framework, the degree of belonging of a given building typology to multiple vulnerability classes. The adoption of the peak ground acceleration for characterising the ground shaking is a further advantage of this study, overcoming several limitations related to the use of macroseismic intensity. [ABSTRACT FROM AUTHOR]