Application Of Time Series Analysis In Forecasting Short Cycle High Incidence Events

Objective To explore the forecasting method for events with quick onset and rapidly-increasing incidence. Methods Using the time series analysis of cucumber downy mildew disease, to explore the forecasting method from the angle of methodology. The current study established ARIMA(2,2,0) model, one-parameter double exponential smoothing model, Holt-Winters two-parameter double exponential smoothing model, ) infectious disease model-based autoregression model, principal component regression model with time factor AR(2)-EGARCH(0,2) model. Results The obtained ARIMA(2,2,0) model is a satisfactory 1-dimension time series model and AR(2)~EGARCH(0,2) is a satisfactory anamorphosis of Autoregressive Conditional Heteroskedasticity (ARCH) model. Conclusion ARIMA(2,2,0) model and AR(2)-EGARCH(0,2) model are suitable for the forecasting of middle stage and late stage of cucumber downy mildew disease. Exponential smoothing model, infectious disease-based autoregression model and principal component regression model can be used as a complementary to the above 2 models for the forecasting of early stage of cucumber downy mildew disease.