CPFR组合预测方法研究

Research on Combination Forecasting Method of CPFR

供应链整合解决方案—CPFR(联合计划、预测、补货)已逐渐被全球许多著名企业采纳并实施,成为供应链减少成本与开拓利润的主要工具.而其最核心部分—联合预测流程的研究还处于探索研究阶段,在简单介绍CPFR的原理和应用现状的基础上,主要讨论CPFR的联合预测流程.从联合预测流程分析着手,创新地引入组合预测方法,提出了基于组合预测模型的联合预测流程,讨论了不同形式的组合预测模型,通过对各预测模型的挑选和组合,减少联合预测流程中因模型的差异而产生的预测异常,为预测异常的处理提供了科学决策的模型.并结合具有良好收缩性能的斯坦规则对组合预测的模型进行校正,建立组合校正预测模型,提高了CPFR联合预测的精度和预测稳定性.

The CPFR （Collaborative Planning, Forecasting and Replenishment）, which is a solution for retail supply chain integration, is widely used and implemented by many world renowned retailers and manufacturers, and has become the irresistible tide of the day for retail supply chain to reduce costs and increase profits. Based on the brief review of the CPFR process, the collaborative forecasting process which is the core of CPFR will mainly discussed. Starting with analysis of the collaborative forecasting process, the combination forecasting method is newly introduced , and CPFR process based-on combination forecasting is proposed and different forms of combination forecasting models also have been studied. By selection and combination operation among forecasting models, discrepancies caused by difference forecasting models are reduced. Combination forecasting provides sdentific decision-making models for dealing with forecasting discrepancies. Taking advantage of good constriction performance of Stein-rule, combination forecasting model is constructed to improve forecasting accuracy and stability of collaborative forecasting models in CPFR process.