零售4.0时代先行发货问题的大数据分析-《长安大学学报（社科版）》

文章信息/Info

Title:: Big data analysis on the issue of dispatch in advance in the retail 4.0 era

Author(s):: MA Lina; WAN Meiyu; DU Yushen; School of Business ，Jilin University，Changchun 130012，Jilin，China

Keywords:: retail 4.0 era; omnichannel retail; big data; supplychain; dispatch in advance; Genetic Algorithm; Association Rule Mining

摘要:: 在零售4.0时代，渠道的多样化不仅丰富了数据源，还能迅速生成大量数据，需要通过分析大数据，为决策提取有意义的信息，通过分析先行发货的重要性，提出了一种基于遗传算法（GA）的优化模型，预测顾客何时购买，然后在顾客线下单前将产品运送到距顾客最近的配送中心，解决先行发货中存在的问题。研究认为，需要先部署云计算来存储所有渠道生成的大数据，再应用基于集群的关联规则挖掘研究顾客的购买行为，根据“如果-那么”预测规则预测未来的采购情况，最后利用修正的遗传算法生成最优的先行发货计划；这种遗传算法考虑了其在运输成本和运输距离之外，还有预测规则的置信度，利用大量的数值实验权衡了先行发货中的不同因素，验证了模型的最优可靠性

Abstract:: In the retail 4.0 era, the diversification of retail sales channels not only enriches the data source, but also can generate huge amounts of data. As the big data needs to be analyzed to provide meaningful information for decision making, this paper proposes an optimized model based on the Genetic Algorithm (GA) through analyzing the importance of dispatch in advance. The model predicts when customers will make the purchase, and dispatches the products to the nearest distribution center to the customer before they make the order online, in order to solve the issues in dispatch in advance. The research finds that cloud computing should be deployed in advance to store the big data generated through all retail sales channels, then the Association Rule Mining based on clusters should be applied to study the purchase behaviors of the customers, and their future purchases will be predicted according to the if-then prediction rule, and finally the revised Genetic Algorithm is used to produce the optimized plan for dispatch in advance. This type of Genetic Algorithm takes the credibility of the prediction rule into account, in addition to the transportation cost and transportation distance, and utilizes a large number of numerical experiments to measure and weigh various factors in dispatch in advance, thus verifying the reliability of the optimized model.