数据驱动的大型网上超市快消品组合策略与定价方法

The problem of bundling and pricing FMCGs (fast moving consumer goods) for online supermarkets has the features of complex customer purchasing modes, complicated relationships between multiple products, and uncertainties between supply and demand. The combination of data mining, data analytics and optimization is investigated to solve this complex decision problem scientifically, practically and precisely. This project will conduct researches from the following aspects. 1) By using the idea of “clustering-filtering-optimization”, the method for generating bundling strategies of FMCGs is investigated. 2) A graph-based demand-forecasting model is constructed. 3) By the method of “clustering-classification”, the uncertainty between the supply and demand is obtained and represented, based on which the data-driven bundle-pricing model is constructed and solved. 4) The application research is carried for the online FMCGs sell platform of Dalian Tiangou..This research is the exploration of using data-driven concept to solve complex decision problems. It is beneficial for the blending and syncretizing of methods of data mining, data analyzing and optimization. The proposed data-driven method for generating bundling strategies can extend the application of data mining in precision marketing. The data-driven bundle-pricing method increase the practicability of optimization. This project is significant for researchers and practitioners because it enriches the directions of e-commerce research and provides guidelines for e-tailers to improve profitability.

大型网上超市快消品组合策略生成与定价具有顾客购买模式复杂、多商品关联关系复杂、供需不确定等特征。针对这一复杂决策难题，融合数据挖掘、数据分析和优化方法，以提高组合策略生成和定价方法的科学性、实用性和精准性为目标，研究以下内容：1)采用“聚类—过滤—优化”思想，研究大型网上超市快消品组合策略生成方法；2)构建基于图模型的快消品组合需求预测模型；3)利用“聚类—分类”手段，获取和表示供需不确定性，构建数据驱动的快消品组合定价模型并求解；4)结合大连天狗网快消品在线销售业务开展应用研究。.本项目为数据驱动求解复杂决策问题开展了有益的探索，有利于促进数据挖掘、数据分析和优化方法的交叉与渗透。提出的数据驱动的快消品组合策略生成方法拓展了数据挖掘方法在精准营销领域的应用；建立的数据驱动的快消品组合定价方法提高了优化方法的实用性。本项目对促进电子商务理论的发展，提高电子商务企业的盈利能力有重要意义。

大型网上超市快消品组合策略生成与定价具有顾客购买模式复杂、多商品关联关系复杂、供需不确定等特征。针对这一复杂决策难题，融合数据挖掘、数据分析和优化方法，以提高组合策略生成和定价方法的科学性、实用性和精准性为目标，重点研究了大型网上超市快消品组合策略生成方法、需求预测模型和定价方法，并结合沈阳京东的实例展开应用研究。主要研究成果有：1)提出基于“聚类—过滤—优化”的快消品组合生成方法，实现了海量数据中的组合策略优选；2)构建基于图模型的快消品组合需求预测模型，实现了商品组合的精准推荐；3)利用“聚类—分类”手段，获取和表示供需不确定性，构建数据驱动的快消品组合定价模型并求解，提高了组合定价的科学性。本项目为数据驱动求解复杂决策问题开展了有益探索，有利于促进数据挖掘、数据分析和优化方法的交叉与渗透，拓展了数据挖掘方法在精准营销领域的应用。本项目对促进电子商务理论的发展，提高电子商务企业的盈利能力有重要意义。