摘要
目的研究季节性差分自回归滑动平均模型(SARIMA)拟合季节时间序列的方法,并将其应用于预测香港流行性腮腺炎疫情趋势。方法利用R软件对2000年1月至2012年8月香港流行性腮腺炎月发病例数资料进行建模,用所构建模型进行预测分析。结果流行性腮腺炎发病呈上升趋势,SARIMA(2,1,1)×(1,1,1)12模型较好地拟合了香港流行性腮腺炎的月发病例数,模型残差为白噪声序列,回代考核平均相对误差为17.5%;后8个月的数据作为前瞻性预测考核,平均相对误差为16.4%。结论 SARIMA模型较好地模拟腮腺炎的流行特征,并进行中、短期预测。
Objective To study the application of SARIMA model in the analysis of mumps epidemic trends in Hong Kong. Method R software was used in the construction of SARIMA model, and the model was then used in the analyzis of the mumps epidemic data obtained from Hong Kong. Results The SARIMA （2, 1,1 ） ×（ 1, 1, 1 ） 12 model was found to be suitable in the simulation of mumps epidemic in Hong Kong. The residual was white noise sequence. The average relative error was 17.5% of back substitution assessment, and 16.4% of the forward-looking forecast assessment using the after eight-month data. It showed a rising trend in the number of incidence of mumps throughout the year in 2012 in Hong Kong. Conclusions SARIMA fitted well with the trends of mumps. The authors explored methodology issues on the prediction of mumps.
出处
《热带医学杂志》
CAS
2013年第1期88-90,106,共4页
Journal of Tropical Medicine
基金
南方医科大学公共卫生与热带医学学院院长基金(GW200821)
