Application of ARIMA and hybrid ARIMA-GRNN models in forecasting AIDS incidence in China

Objective To explore a suitable model in forecasting monthly incidence of AIDS in China and offer some references for AIDS prevention. Methods The data was collected from January 2011 to December 2017 to build the autoregressive integrated moving average (ARIMA) model and hybrid ARIMA-generalized regression neural network (GRNN) model. The data from January to May 2018 was used to evaluate two models' forecasting performance. Result A strong seasonal variation can be seen and ARIMA(1,1,1) (0,1,1)₁₂ was selected as the most suitable model to forecast the AIDS incidence in China. The smooth factor of hybrid model was 0.021. The hybrid ARIMA-GRNN model had better fitting and forecasting performances than single ARIMA model. Conclusion Both of the two models were suitable in forecasting monthly incidence of AIDS and the hybrid model was better than single ARIMA model.