摘要
人工议价是当前网络货运平台的主要定价方式,但存在议价回合多、耗时及效率低下等弊端。为了弥补人工议价的不足,本文提出基于深度学习的动态价格预测模型。首先,结合历史订单数据和极端随机树特征选择的结果,对价格预测模型进行训练调优。其次,利用该模型对待定价订单的计划价格和实际价格进行预测,形成一个价格区间,制定实时价格策略,为网络货运平台的定价决策提供依据。最后,论文以顺丰网络货运平台为例,选取了8种经典机器学习模型作为对比模型,经对比发现本文所建的模型R2-score达到98.24%,相较于与其他模型有更高的预测准确度,并能为平台提供多样化的动态调价策略。实验结果表明:该深度学习模型定价预测更精准、动态定价能力更强,能够适应复杂的现实交易情况,从而可提高网络货运平台在公路货运市场的竞争力。
Manual bargaining is currently the main pricing method for online freight platforms,but there are drawbacks including multiple rounds of bargaining,and time-consuming and inefficient negotiations.To compensate for the shortcomings of manual bargaining,this paper proposes a dynamic price prediction model based on deep learning.First,the price prediction model is trained and optimized using historical order data and the results of extreme random tree feature selection.Next,the model is used to predict the planned and practical prices of pending orders,formulating a price range and establishing a real-time pricing strategy to provide a basis for pricing decisions on the online freight platform.Finally,the SF Express online freight platform is considered as a case study and 8 classic machine learning models are selected for comparison.As a result,the R2-score of the model proposed in this paper reaches 98.24%,with higher prediction accuracy than other models,and the model can provide diversified dynamic pricing strategies for the platform.Further,experimental re-sults show that the deep learning model has more accurate pricing prediction and stronger dynamic pricing capabilities,which can adapt to complex real-world transaction situations and improve the competitiveness of the online freight platform in the road freight market.
作者
姜浩晨
张小强
JIANG Hao-chen;ZHANG Xiao-qiang(School of Transportation and Logistics,Southwest Jiaotong University,Chengdu 611756,China;National Engineering Laboratory of Integrated Transportation Big Data Application Technology,Chengdu 611756,China;National United Engineering Laboratory of Integrated and Intelligent Transportation,Chengdu 611756,China)
出处
《交通运输工程与信息学报》
2023年第3期86-97,共12页
Journal of Transportation Engineering and Information
基金
中国铁路南宁局集团有限公司科研计划项目(KYL202112-0200)。
关键词
物流工程
动态定价
深度学习
网络货运平台
logistics engineering
dynamic pricing
deep learning
online freight platform
