南通市水痘疫情时间序列分析

Time series analysis of the epidemic situation of varicella in Nantong City

目的应用季节乘积求和自回归移动平均模型分析南通市水痘每月发病数时间序列,为建立疫情预测模型提供参考依据。方法利用Eviews软件对南通市2008至2014年间水痘病例月报告数进行建模,首先采用自然对数转换、差分法平稳序列,然后估计模型参数,对模型进行检验,筛选最优模型,最后进行预测分析。结果成功建立模型ARIMA(1,0,0)(1,1,0)12,模型表达式为:(1-B12)lnxt(1-0.579 4B)(1+0.312 2B12)=t。模型通过残差白噪声检验,P>0.05;平均绝对百分误差为8.90%,预测精度较高。运用该模型预测2015年1—6月病例数分别为67,36,64,55,72和87例。结论该模型对南通市水痘发病情况拟合和趋势预测结果较好,可根据预测结果开展水痘疫情相关防控准备工作。

Objective The purpose of this study was to establish the predictive model by using SARIMA model to analyze monthly data of varicella cases in Nantong city. Methods Eviews software was used to fit SARIMA model for the monthly data of varicella cases in Nantong city. Firstly, the natural logarithm transformation and difference method was used to make the data se- quence stationary. Secondly, parameters of model were estimated and tested. Meantime, the optimal model was selected. Finally, the optimal model was used to forecast and analyze. Results The model ARIMA （ 1,0,0 ） （ 1,1,0 ） 12 was established and indicated by was （ 1-B12） lnxt（ 1-0.579 4B） 1 1＋0.312 2B12 ）= et,which had passed the white noise test（ P〉0.05）. All values were accurately fitted with MAPE being 8.90%. Accordingly, the model was applied to predict the monthly numbers of varicella cases during the first half of 2015,which were 67,36,64,55,72 and 87 respectively. Conclusion The model is better fitted with the time series of varicelia cases in Nantong city. Relevant preparatory prevention measures should be taken against the varicella outbreak according to the forecast results.