摘要
目的探讨时间序列自回归移动求和平均模型在时间序列资料建模和预测中的应用,建立全国麻疹发病率的预测模型,为麻疹的预防控制提供理论依据。方法利用1950—2012年麻疹发病率建立时间序列模型,利用2013一2014年数据验证预测模型的合理性,对2015—2017年麻疹发病率进行预测。结果单位根检验结果表明,经对数变换后二阶差分数据ADF_T小于其他不同检验水平的临界值(P<0.001),数据平稳;根据最大似然估计建立ARIMA(0,2,1)模型;模型残差满足白噪声过程,模型拟合较好;对2015—2017年麻疹发病率的预测值分别为1.33/10万,5.36/10万,2.17/10万。结论 ARIMA能较好的拟合全国麻疹发病率的变动趋势,并能对短期内麻疹发病率进行较高精度的预测,根据预测结果,我国麻疹发病率在2017年出现下降的拐点。
[Objective]To discuss the application of time series Autoregressive Integrated Moving Average Model（ARIMA） in time series modeling and prediction,establish the prediction model of national measles incidence,and provide theoretical basis for measles prevention and control.[Methods]Time series model was established according to measles incidence in 1950-2012,was validated according to incidence in in 2013-2014,and was used to predict the unknown data in 2015-2017.[Results]The results of unit root test showed that the second order differential data after logarithmic transformation ADF_T was smaller than the critical value（P 0.001） of every other test levels,and the data was stationary. ARIMA（0,2,1）model was obtained according to the maximum likelihood estimation.The model residuals met the white noise process,and the models fit well. The prediction of measles incidence in 2015-2017 was 1.33/lakh,5.36/lakh,and 2.17/lakh,respectively.[Conclusion]ARIMA can fit the trend of measles incidence in China perfectly and can predict the incidence of measles accurately in the short term. According to the model predictions,the incidence of measles in 2017 will be a turning point in China. incidence of measles in 2017 will be a turning point in China.
作者
陈丽萍
王小万
李建
CHEN Li-ping;WANG Xiao-wan;LI Jian(Institute of Medical Information,Peking Union Medical College ,Beijjing, 100020,China;a Health Policy and Management Research Center,b Health Economics Research Room,Institute of Medical Information,Chinese Academy of Medical Sciences, Beijing, 100020, China)
出处
《职业与健康》
CAS
2018年第9期1203-1207,共5页
Occupation and Health
基金
中央级公益性科研院所基本科研业务费项目(2016ZX330028)
