自回归综合移动平均模型对天津市甲型肝炎发病预测

Prediction of the incidence of hepatitis A in Tianjin using the autoregressive integrated moving average model

摘要: 目的　用自回归综合移动平均模型（ARIMA）季节乘积模型（p，d，q）（P，D，Q）s对天津市甲型肝炎（甲肝）发病资料建模并预测，评价模型的预测效果。方法　通过对差分方法使原始序列平稳，依据AIC和SBC准则确定模型阶数，采用条件最小二乘方法估计模型参数，最终建立起ARIMA预测模型。结果　对甲肝数据建立了乘积ARIMA（2，1，1）（0，1，1）12模型，预测误差为3.72%。结论　ARIMA是一种短期预测精度较高的预测模型。

Abstract: Objective　The study established a predictive model of multiple seasonal autoregressive integrated moving average (ARIMA) (p, d, q)(P, D, Q)s based on the hepatitis A data, and evaluated its predictive effects. Methods　The primitive series stabilized using the finite difference method, the order of model confirmed according to the Akaike Information Criterion and Schwarz Bayesian Criterion, and the parameters of model obtained through conditional least squares, the ARIMA predictive model was established. Results　The error of the multiple ARIMA(2, 1, 1)(0, 1,1)12 model for the prediction of hepatitis A was 3.72%. Conclusion　The ARIMA was a highly accurate model for short-term prediction of incidence of hepatitis A.