Application of Autoregressive Integrated Moving Average Model in Forecasting of the Monthly Incidence Rate in Baoji

YANG Pei-rong,TIAN Hui,YAN Chuan-yuan
2012-01-01
Abstract:Objective To set up a prediction model of the monthly incidence rate of bacillary dysentery in Baoji districts by applying autoregressive integrated moving average(ARIMA) model,and try to quantitatively reflect the epidemic characteristic of bacillary dysentery in Baoji districts.Methods ARIMA model was established based on the monthly incidence rate of bacillary dysentery of Baoji from January 2004 to December 2007.The monthly incidence rate of bacillary dysentery of Baoji from January 2004 to June 2010 was used to forecast the monthly incidence rate during July 2010 to December 2010.The forecast result was also assessed.Results The model of the monthly incidence rate of bacillary dysentery in Baoji was ARIMA(1,0,0)(0,1,1)12,in which the Auto-regressive(AR1) was 0.605(t=6.030,P0.001);the Seasonal Moving-Average(SMAl) was 0.779(t=3.471,P0.001);the statistics of Ljung-Box Q has no statistical significance(Ljung-Box Q(18)=8.503,P=0.932);and the residuals was the white noise,The mathematic function was: lnYt=0.605lnYt-1+0.799lnYt-12-0.12.The predicted monthly incidence ^rate of dysentery from July to December in 2010 consisted with the actual one,with the maximum absolute value of absolute error as 1.896,the minimum absolute value,0.039;the relative error,6.78%.There was little difference between the predicted value and the actual value,and the predictive effect of the model was good.Conclusions The ARIMA model can be used to reflect epidemic characteristic of bacillary dysentery in Baoji districts,and to forecast the future incidence rate.The model can predict the trends of the monthly incidence rate of bacillary dysentery.
What problem does this paper attempt to address?