摘要
目的探讨时间序列分析中自回归移动平均模型在六安市细菌性痢疾发病预测的可行性和适用性,为早期做好防控工作提供科学依据。方法使用SPSS 17.0软件对六安市2003年1月～2012年12月的细菌性痢疾月发病率建立ARIMA模型,以2013年的1～7月实际发病率作为预测模型的考核样本,验证模型的预测效果。结果六安市细菌性痢疾月发病率模型为ARIMA(0,0,1)×(0,1,1)12,模型移动平均参数MA1=-0.473(t=-5.153,P<0.05),季节移动平均参数SMA1=0.937(t=2.494,P=0.014);残差分析Ljung-BoxQ统计量经检验,差异无统计学意义(Ljung-BoxQ=10.208,P=0.856),提示残差为白噪声。模型预测的平均相对误差为27.82%,但预测的动态趋势与实际值基本吻合,且实际值均在预测值的95%可信区间内。结论 ARIMA(0,0,1)×(0,1,1)12模型可为六安市细菌性痢疾的防控提供参考。
Objective To explore the feasibility of application of autoregressive integrated moving average （ ARIMA ） model to forecast the incidence rate of bacterial dysentery in Lulm city, and provide scientific basis for early prevention and control of bacterial dysentery. Methods ARIMA model was established by using SPSS 17.0 software based on the monthly incidence rate of bacillary dysentery of Lu＇an from January 2003 to December 2012. The actual monthly incidence rate of bacillary dysentery from January to July 2013 was used as inspection sample to predict the model, and the forecast result was also assessed. Results The model of monthly incidence of bacillary dysentery in Lu＇an was ARIMA （0,0,1） × （ 0,1,1 ） 12, in which moving average （ MA1 ） was - 0. 473 （ t = - 5. 153, P 〈 0.05 ） and seasonal moving average （ SMA1 ） was 0. 937 （ t = 2. 494,P 〈 0.05 ） , Ljung - Box Q had no statistical significance （ Ljung - BoxQ = 10.208,P = 0.856 ） , and residuals was the white noise. The average of the relative error between actual and predicted values was 27.82% , but the dynamic trend of model prediction and the actual value was basically same, and actual values were within the predictive value of 95 % confidence interval. Conclusion The ARIMA （ 0,0,1 ） × （ 0,1,1 ） 12 model could provide a reference to the prevention and control of bacillary dysentery in Lu＇an city.
出处
《安徽预防医学杂志》
2014年第3期175-177,180,共4页
Anhui Journal of Preventive Medicine
