Abstract

A naive Bayes classifier is determined to predict intensities from peak ground velocity and acceleration. It is trained on the same dataset that was used in the study of Faenza and Michelini (2010). The naive Bayes classifier directly estimates a discrete probability distribution for the ordinal intensities. Comparisons based on generalization error, estimated by cross-validation, show that the naive Bayes classifier performs better than traditionally employed regression models.

You do not currently have access to this article.