基于时间序列和灰色模型的交通事故预测

Traffic Accidents Prediction Based on Time Series and Grey Model

利用时间序列和灰色模型理论,针对北方某城市的交通事故统计数据,分别建立了自回归移动平均模型及灰色模型,并对各模型进行了步长为12的预测。通过模型对比发现:2个模型的预测绝对误差分别为23.95%和54.32%,且对于具有季节周期性特点的序列,自回归移动平均模型的预测结果与实际观测值比较吻合,说明自回归移动平均模型比灰色模型更能充分挖掘历史信息以减少预测误差,并反映数据的周期性变化,具有良好的适用性。

According to the time series model theory and grey model,auto regression moving average model and grey model prediction model are separately built based on the statistics data of road traffic accidents collected from some cities in northern China.Both of the models forecast 12 step lengths.The comparison of models shows that the mean absolute percentage errors of the two models are respectively 23.95% and 54.32%,and auto regression moving average model prediction agrees well with the series recognized as seasonal movements or variations.Therefore,it is concluded that time series model is more applicable for history data mining in minimizing prediction error than grey model,and can reflect the cyclic movements precisely.