基于ARIMA-GM组合模型的邮电业务总量预测

对传统预测具有波动性及季节性双重趋势时间序列的模型—ARIMA乘积季节模型进行了改进,先用ARIMA乘积季节模型对邮电业务总量历史数据进行识别和拟合,然后用GM(1,1)模型对其带阀值的残差序列进行修正,最后结合二者得到ARIMA-GM这一组合预测模型.利用此模型对09年上半年中国邮电业务总量进行了预测,结果表明,组合预测方法比单项ARIMA乘积季节模型预测具有更高的精度.

Forecast of Business Volume of Past and Telecommunications Based on Arima-GM Combined Model

Aim at the time series which have fluidity and seasonal double trend,this paper improved the traditional forecast technique—ARIMA product season model.Firstly,used the ARIMA product season model to distinguish and fit the historical business volume of P& T data.Secondly,used the GM(1,1) model to amend the residual sequence with threshold. Finally,obtained the ARIMA-GM model through combining the two above.This paper used this model to forecast the China business volume of P & T in the first half year of 2009. Forecast results indicated that the combined forecast model approach enjoyed more percise forecast than monomial ARIMA product season model forecast approach.