自回归移动平均混合模型在中国道路交通伤害预测中的应用

Autoregressive integrated moving average model in predicting road traffic injury in China

探讨时间序列分析的自回归移动平均混合模型（ARIMA）在中国道路交通伤害（RTI）预测中的应用。收集1951—2011年中国道路交通伤害资料，进行时间序列分析，建立ARIMA模型。构建得到RTI事故起数ARIMA（1，1，0）预测模型为Yt=e^Yt-1＋0.456 Yt-1＋et,其中et为随机误差，模型残差序列为白噪声，Ljung．Box检验P〉0．05，统计量无统计学意义，拟合效果良好。应用该模型预测2011年中国RTI事故起数，预测值与实际观测结果相符，实际观测值在预测值95％CI内。用该模型预测2012年中国RTI事故起数，预测值（95％凹）为207838（107579～401536）。应用ARIMA模型能较好地预测中国道路交通伤害情况。

This research aimed to explore the application of autoregressive integrated moving average （ARIMA） model of time series analysis in predicting road traffic injury （RTI） in China and to provide scientific evidence for the prevention and control of RTI. Database was created based on the data collected from monitoring sites in China from 1951 to 2011. The ARIMA model was made. Then it was used to predict RTI in 2012. The ARIMA model of the RTI cases was Yt=e^Yt-1＋0.456 Yt-1＋et, stands for random error）. The residual error with 16 lags was white noise and the Ljung-Box test statistic for the model was no statistical significance. The model fitted the data well. True value of RTI cases in 2011 was within 95%CI of predicted values obtained from present model. The model was used to predict value of RTI cases in 2012, and the predictor （95%CI） was 207 838 （ 107 579-401 536）. The ARIMA model could fit the trend of RTI in China.