基于灰色关联分析的季节性民航货运量预测

Study on seasonal air freight volume prediction based on grey correlation analysis

为更精确地预测月度航空货运量,提出组合模型预测方法.该模型由季节GM(1,1)和季节ARIMA乘积模型构成,它结合了该2种模型中时间序列预测的优点.灰色模型GM(1,1)能准确反映时间序列的增长趋势;ARIMA乘积模型对季节特征有较好的拟合效果.依据霍尔特温特预测模型计算季节性GM(1,1)模型的季节指数,并用灰色关联分析求出组合预测中的权值.组合预测模型的平均相对误差为0.62%,而季节性GM(1,1)模型和ARIMA乘积模型的平均相对误差分别为4.49%和-3.16%.预测分析结果说明,该模型的非线性曲线拟合精度和预测精度明显高于单个模型,可较好地反映系统的动态性和运量的季节时序关联性,为季节性时间序列预测提供了新的途径.

In order to forecast the monthly freight volume of civil aviation accurately,this paper proposed a combined model.This model is composed of the multiple seasonal time series autoregressive integrated moving average model（SARIMA） and grey theory model.Also it has the advantages of time series prediction.GM（1,1） model can reflect the increasing trend of the time-series,while the multiple seasonal autoregressive integrated moving average model has better fitting results in seasonal characteristics.And calculates the seasonal index according to holt-winter forecasting model,and obtained the combination prediction weights with the grey related theory.An average relative error of the combined model is 0.62%,wihle the index of seasonal GM（1,1） and multiple SARIMA are 4.49% and-3.16% respectively.The model of non-linear curve fitting and prediction precision is significantly higher than that of a single model.It can better reflect the dynamic characteristic of system and the seasonal time series relevancy for monthly freight volume.Thus,it provides a new way for the seasonal time series prediction.