季节性差分自回归滑动平均模型在上海市道路交通伤害预测中的应用

[目的]探讨季节性差分自回归滑动平均(SARIMA)模型预测道路交通伤害的可行性,为掌握上海市交通伤害趋势提供依据。[方法]利用EVIEWS软件对2000—2009年上海市道路交通伤害死亡的季度数据进行SARIMA模型拟合,并利用2010年数据对预测数据进行验证。[结果]上海市道路交通死亡具有明显的季节要素,趋势要素呈逐步下降趋势;对原始图形识别后,综合考察几种模型拟合优劣,最终采用SARIMA(2,1,0)(0,1,1)4,其能很好地拟合上海市道路交通伤害死亡情况。2010年4个季度死亡率预测值分别为1.49/105、1.74/105、1.93/105和2.06/105,实际值均在预测区间内,残差也显示为白噪声序列。预测结果较好。[结论]SARIMA模型是一种能较好地预测道路交通伤害趋势的工具,可为预防与控制道路交通伤害提供决策依据。

Application of Seasonal Autoregressive Integrated Moving Average Model to Road Traffic Injury Prediction in Shanghai

To explore the feasibility of seasonal autoregressive integrated moving average （SARIMA） model in predicting road traffic injury, and to provide reference for road traffic injury trends in Shanghai. [ Methods ] A SARIMA model was presented to fit the seasonal road traffic mortality data of Shanghai （2000-2009） via EVIEWS software, and estimated mortalities of 2010 were verified with the actual data. [ Results ] The seasonal component was statistically significant in Shanghai＇s road traffic mortality data. A decreasing trend was observed in the trend component of the model. SARIMA （2, 1, 0） （0, 1, 1）4 was the best fitting model among various candidate models. The predicted seasonal mortalities of 2010 were 1.49/105, 1.74/105, 1.93/105, and 2.06/105 respectively. The actual values were all in the prediction intervals, and the residuals were considered,as white noise serial. The verification with actual data passed our test. [ Conclusion ] A SARIMA model can be used in accurate trends prediction of road traffic injury and therefore can provide evidences for road traffic injury intervention.