An approach of Bayesian networks in magnitude forecast based on earthquake trace cloud
Xiao Fan, Yong Guang Zhao, Shoudong Han · 2012
Magnitude forecast is an indispensable part of the earthquake forecast. In order to get better predicting results, this paper expounds the theory of earthquake trace cloud, and introduces the Bayesian network method to improve it and develop the automated processing. Firstly, the paper selects the proper variables after analysing the features of earthquake trace cloud and the practical situation. Then the paper completes the Bayesian learning to build the Bayesian network model of magnitude forecast. Finally the paper uses the junction tree algorithm to do Bayesian inferences to predict magnitude. The experimental results indicate that Bayesian network is effective in magnitude forecast, and it has advantages over the manual network and Naive Bayesian network.