Debris flow prediction research based on two-dimension Bayesian classifier

Debris flow is one of the most dangerous natural disasters, so correct prediction of debris flow is very important. In this paper, a two-dimension Bayesian classifier, using rainfall data, is proposed to predict the debris flow. Firstly, part of historical rainfall and debris data are used to learn and train for classifier, which is a two-dimension classifier, then daily rainfall and recent five-day rainfall are used as the two dimensions inputs, on the basis of which the parameters of the classifier are confirmed; at last, in order to test the Bayesian classifier, the historical rainfall data is used for predicting debris flow, the predicted results are compared with actual results of debris flow occurrence, and the accuracy of classifier can be computed. The experiment shows that the two-dimension Bayesian classifier can obtain more correct prediction than one-dimension classifier, with the accuracy reaching 88.5%.