关键词: 物流需求, 主成分分析, 包容性检验, 组合模型

Abstract: Model selection and how to combine is the key of logistics demand forecasting. In order to improve the forecasting accuracy of logistics demand, a logistics demand forecasting model is proposed based on Encompassing Tests and Principal Component Analysis（ET-PCA）. Some single models are used to forecast the logistics demand, and then encompassing tests are used to select good single model. The selected single models are combined by PCA, and the simulation experiment is carried out to test the performance of the combination model on logistics demand data. The simulation results show that, compared with traditional combination model, ET-PCA has solved the problem of how to select  and combine the forecasting models. It can be more comprehensive, accurately describe the change trend of logistics demand, and improve the prediction accuracy of logistics demand. It has a good application value.

Key words: logistics demand, Principal Component Analysis（PCA）, encompassing tests, combination model