基于XGBoost的卷烟送货时间预测仿真模型研究

Research on Simulation Model of Cigarette Delivery Time Prediction Based on XGBoost

为节约路线更改所花费的时间、降低路线调整工作量、提高客户服务满意度。提出了一种用于估计卷烟送货时间的仿真模型,该模型基于离散事件系统仿真法,以湖北省十堰市烟草公司配送任务为例,通过XGBoost机器学习模型进行送货时间预测仿真验证。实验结果显示论文设计的学习模型在配送路线整体时间的预测上可以实现92.11%的准确率,在客户的预计送达时间预测上准确率高于80%,预计送达时间预测误差小于半小时。通过实际案例验证了模型的科学性、准确性和有效性,为卷烟物流路线的动态改变提供了一个有效解决方案和实用辅助工具。

In order to reduce the time cost of the route adjustment process,reduce the workload of the route adjustment and improve the service satisfaction of the retail customers,a predict simulation model of cigarette delivery time is proposed,based on the simulation method of discrete event system and using the Shiyan City Tobacco Company,Hubei Province as example,through XGBoost machine learning model to do the predict simverification of delivery route time.The results show that the predict accuracy of delivery time for the whole route reaches 92.11%,the predict accuracy of delivery time for a single retail customer exceeds 80%,and the average delivery time is accurate to within half an hour.The accuracy and the scientificity of the model are verified through practical cases,and an effective auxiliary tool is provided for the optimization and adjustment of cigarette logistics route.