ARIMA模型在北京市西城区细菌性痢疾发病预测中的应用  

Application of ARIMA Model for predicting the epidemic situation of bacillary dysentery in Xicheng District of Beijing

作　　者：关静[1] 张燕[1] 宋静[1] 

机构地区：[1]北京市西城区疾病预防控制中心,100120

出　　处：《职业与健康》2015年第23期3243-3245,3248,共4页Occupation and Health

摘　　要：目的尝试构建适用于北京市西城区细菌性痢疾发病特点的统计预测模型,为该区传染病定量预测的实施做出探索。方法应用SPSS 13.0统计软件对北京市西城区2004年1月-2013年12月细菌性痢疾逐月发病情况进行求和自回归滑动平均模型(ARIMA)建模和拟合,利用筛选出的最优模型对2014年1-12月的发病情况进行预测,并评价模型的预测效果。结果筛选的最优乘积模型为ARIMA(1,0,0)(0,1,1)_(12),BIC=27.426,模型拟合效果的度量Box-Ljung Q差异无统计学意义(Q=10.949,P=0.813),模型残差序列为白噪声。模型预测值与实际值拟合较好,实际值均在预测值95%可信区间范围内。结论 ARIMA模型能够应用于北京市西城区细菌性痢疾流行趋势的预测,为实施干预措施提供科学依据。[Objective]To try to establish a statistical model for predicting the epidemic characteristics of bacillary dysentery in Xicheng District of Beijing,explore the implementation of quantitative prediction of infectious diseases in this district.[Methods]By using the SPSS 13.0 statistical software,the modeling and fitting of autoregressive integrated moving average（ARIMA） were performed according to monthly epidemic situation of bacillary dysentery from January 2004 to December 2013.The optimal model was used to predict the epidemic situation from January to December in 2014,and the prediction effect was evaluated.[Results]The optimal model is ARIMA（1,0,0）（0,l,l）_（12）,BIC=27.426,Box-Ljung Q test showed that there was no statistically significant difference in fitting effect（Q=10.949,P=0.813）,and the residual series is white noise.The predicted values agree well with the actual value,and all actual values fell in the 95%confidence interval of predictive value.[Conclusion]ARIMA model can be used to predict the epidemic situation of bacillary dysentery in Xicheng District of Beijing,and provide scientific basis for carrying out the intervention measures.

关 键 词：求和自回归滑动平均模型 细菌性痢疾 预测 

分 类 号：R181.2[医药卫生—流行病学]