浙江省细菌性痢疾月发病率ARIMA模型建立及预测分析

目的构建ARIMA模型预测浙江省细菌性痢疾的月发病率。方法利用SAS 9.0统计软件对浙江省2001—2011年2月的细菌性痢疾发病率数据建立ARIMA模型,并进行预测分析。结果拟合ARIMA(1,0,0)12模型的AIC为227.23,为细菌性痢疾的月发病率最佳模型,该模型预测值与实际值的平均相对误差为15.9%,实际值都在95%的可信限之内,预测值与实际值较为接近。结论 ARIMA模型可以较好的预测细菌性痢疾发病率的变化趋势,能够运用于细菌性痢疾发病趋势的预测及预警,为防控措施的制定提供参考。

Application of ARIMA Model for Estimating the Incidence of Bacillary Dysentery

Objective To establish an ARIMA model for the prediction of the monthly incidence of bacillary dysentery in Zhejiang province. Methods The ARIMA model was established by using monthly incidence data of bacillary dysentery in Zhejiang province from January, 2001 to February, 2011. In order to test the effectiveness of the ARIMA model, the identified coefficient and AIC （Akaike inforomtion criterion） were comprehensively dealt with to establish the ARIMA model. Results The AIC of the ARIMA （ 1, 0, 0） 12 model was 227.23, which was the fittest model to predict the monthly incidence of bacillary dysentery. The relative error was 15.9% and the real value was included in the 95% CI. The predicting value was close to the true value. Conclusion ARIMA model was fit for predicting the trend of the incidence of bacillary dysentery and early warning of the epidemic.