安徽省滁州市细菌性痢疾发病率ARIMA模型建立与分析

Establishment and analysis on ARIMA model of the incidence of bacillary dysentery in Chuzhou city

目的构建安徽省滁州市细菌性痢疾发病率的ARIMA模型,预测细菌性痢疾发病趋势。方法收集1987~2013年安徽省滁州市细菌性痢疾年发病率资料,通过SPSS软件拟合ARIMA模型,采用最大似然法估计模型参数,依据赤池信息准则与贝叶斯信息准则确定模型的阶数,用Q统计量对模型适应性进行检验,建立ARIMA预测模型。结果通过对参数和模型的拟合优度检验以及残差白噪声序列的检验,最终确定模型为ARIMA（1,1,1）。AIC=5.573,BIC=8.165,统计量Q=8.857〈χ20.05,26。模型预测值与实际值的平均误差率MER=0.338。结论ARIMA模型能够应用于安徽省滁州市细菌性痢疾流行趋势的预测及预警,为实施干预提供依据。

Objective The study was designed to develop the ARIMA models for predicting the incidence of bacillary dysentery in Chuzhou city, and provide supports for prevention and control of the disease. Methods Data of bacillary dys- entery incidence from 1987 to 2013 in Chuzhou city were collected. SPSS was used to fit ARIMA model, parameter of model was estimated based on maximum likelihood. The order of model was determined according to AIC and BIC criterion. Q sta- tistic was used to verify the applicability of the model. Results Through the test of parameters and goodness of fit as well as white - noise residuals , we finalized the model ARIMA （ 1 , 1, 1 ） , of which AIC （Akaike Information Criterion） =5. 2 573, BIC （ Bayesian Information Criterion） = 8. 165, statistic of Q（ 8. 857） was smaller thanx0.05,26. The MER （ Mean Rel- ative Error） of ARIMA model were 0.338. Conclusion The ARIMA model of bacillary dysentery incidences could be applied to forecasting the epidemic trend of bacillary dysentery in Chuzhou city. This also provides a basis for the intervention.