[The application of ARIMA in forecasting the cases of rabies in China different human groups]

Objective: To understand the situation of rabies in China different human groups. Seasonal ARIMA model is established to make forecasts and to provide suggestions for human rabies controlling in China. Methods: Build a model with data during 2004-2013, and validate the model by data of 2014. Then predict the cases of rabies from 2015-2017. Results: The long term trend is decreasing year by year, while seasonal effect is found that number of the third and fourth quarter are higher than others. The seasonal ARIMA model was built, whose residual are white noise. The RMAE in forecasting of peasants, students and scattered children are 19.10%、49.93% and 68.01%. Conclusion: The decreasing of rabies cases in China with time shows that the measures for control are effective. October for peasants, September for students, August for scattered children are critical period in future for rabies' control. Seasonal ARIMA model is a feasible model in forecasting the cases of rabies in China different groups in some way, error will be reduced by modeling separately for different wave sequence and combining with other models like ARIMA-GARCH.