Combined Forecast Model and Application Research of Tobacco Sales Based on Group Method of Data Handling and Auto Regression Integrated Moving Average
Tobacco sales prediction is important to policy formulation and management upgrading of Chinese Tobacco.In this paper,the characteristics and impact factors about tobacco sales forecast system were detailed analyzed.Particularly,aiming at the long-term growth trends and seasonal fluctuations of monthly sales,a hybrid method which combined both ARIMA and GMDH models were proposed.The proposed method took advantage of each models strength in linear and onlinear modeling.Real sales data (Jan 2002 to Mar 2007) of one municipal tobacco commercial company were used to test this model.The tested model was used In sales prediction of first three months 2007.By analyzing the PE (Percentage Error) and MAPE (Mean Absolute Percentage Error) of forecast results,it indicates that the accuracy of combination forecast model can fit the need of forecast process and the proposed method is an effective way to tobacco sales prediction.
Tobacco sales forecast GMDH ARIMA Combined forecast data mining
Weimin Liu Aiyun Zheng Sujian Li Jiangsheng Sun Fanggeng Zhao Zhihong Li
Department of Logistics Engineering,University of Science and Technology Beijing Beijing 100083,Chin Department of Mechanical Engineering,Hebei Polytechnic University ,Tangshan,Hebei Province 063009,Ch Department of Logistics Engineering,University of Science and Technology Beijing Beijing 100083,Chin Department of Industrial Engineering,Tsinghua University,Beijing 100084,China
国际会议
大连
英文
316-323
2008-07-27（万方平台首次上网日期，不代表论文的发表时间）