组合模型在港口集装箱吞吐量预测中的应用

根据某港的集装箱吞吐量的历史数据,分别利用时间序列的三次指数平滑法、改进的GM(1,1)残差模型以及组合预测方法对某港的集装箱吞吐量进行了探讨,并对未来几年某港集装箱吞吐量进行了预测.结果证明:组合预测模型融合了前两种预测方法的优点,其平均绝对误差、平均相对误差绝对值以及误差的标准差分别为4.5208,0.0343,49.5161,都小于单独用三次指数平滑法或改进的GM(1,1)残差模型时的平均绝对误差、平均相对误差绝对值以及误差的标准差,达到最优的精度等级.

Application of combined model in forecasting container throughput capacity of a harbour

The paper analyzed the container throughput capacity of a harbour by using cubic exponential smoothing,improved remnant difference correct model GM(1,1) and combined forecast method,and forecast the container throughputs of the harbour in the next few years.The result implied that by melting the advantages of the two former forecast method,combined forecast model's mean absolute error,mean relative error and standard deviation are 4.5208,0.0343,49.5161 respectively,less than those when using cubic exponential smoothing or improved remnant difference correct model GM(1,1) respectively, thus the combined forecast model obtained the best precision.