基于GBRT树模型分位数回归预测的CPFR补货方法

CPFR Replenishment Method Based On Quantile Regression Prediction of GBRT Tree Model

随着大数据的发展和物流科技信息化进程的加快,企业供应链数据呈爆炸式增长,且种类繁多、关系网络复杂,而传统CPRF技术中的预测模型已经不能适应供应链大数据需求预测,更不能依据需求预测进行有效的库存管理,经典的周期库存盘点策略也不能很好地适应非正态分布的需求数据,如何对供应链大数据进行准确预测并补货已成为供应链研究的热点。依据大数据的分位数回归预测技术,利用历史数据信息进行准确预测,并将分位数回归预测与补货模型合理有效连接,通过真实数据仿真分析,表明在98%的服务水平下,平均库存得到了降低。

With the development of big data and the acceleration of logistics technology informatization,the data of enterprise supply chain has shown explosive growth,and there are many kinds and complex relationship networks.The traditional prediction model in CPRF technology can neither meet the demand prediction of large data in supply chain,nor the effective inventory management based on demand prediction.The classical periodic inventory strategy cannot well adapt to the demand data of non-normal distribution.How to accurately predict and replenish the large data of supply chain has become a hot spot of supply chain research.Based on the quantile regression prediction technology of large data,this paper uses historical data information to predict accurately,and links quantile re?gression prediction with replenishment model reasonably and effectively.The simulation analysis of real data shows that under 98%ser?vice level,the average inventory has been reduced.