顾客导向的目录分割问题研究
发布时间：2018-09-19 12:37
【摘要】：互联网时代,数据量呈爆发式增长,对海量数据中有价值的信息的提取成为一种迫切需求。为了满足这一需求,数据挖掘技术应运而生,并引起了越来越多的学者的关注。在这一背景下,诸多数据挖掘的理论观点被不断提出,基于微观经济的数据挖掘框架就是其中之一。顾客导向的目录分割问题是基于微观经济观点的数据挖掘在商业领域中的一个重要的研究课题,希望通过为企业设计适合的目录发送给相应的顾客,最大化被覆盖的顾客的数量。本文在对顾客导向的目录分割问题深入研究的基础上,提出了新的目录分割问题以及有效的目录分割算法。在电子商务环境下,顾客的购买行为受诸多因素影响,论文基于网上消费者的购买心理和行为,对加入利润约束的目录分割问题进行扩展,提出考虑运费的双目标目录分割问题,以最大化企业期望利润和被覆盖顾客数为目标。此外,基于顾客价值理论,考虑企业长远效益,提出新的感知导向的目录分割问题,以提高目录商品的点击率和转化率。论文提出了一种改进的协同遗传算法,算法将种群分为两个子种群,实现种群内部进化和种群之间的共同进化,并通过最优解集以及自适应的变异算子的设计,使算法的搜索过程跳出局部最优,加速算法的收敛。此外,利用合作交叉算子实现种群之间的信息交互,使得两个子种群互相适应、共同进化。基于真实的电商数据,论文将提出的算法和memetic算法以及经典的RBPF算法做对比分析,结果表明提出的算法能覆盖更多的顾客,优于前两个算法。基于提出算法,论文通过原型系统实现目录定制的可视化。
[Abstract]:In the Internet era, the amount of data increases explosively, so it is an urgent need to extract valuable information from massive data. In order to meet this demand, data mining technology emerges as the times require, and has attracted more and more scholars' attention. Under this background, many theories of data mining have been put forward, and the framework of data mining based on microeconomics is one of them. The problem of customer-oriented catalog segmentation is an important research topic of data mining based on microeconomic viewpoint in the business field. It is hoped that it can be sent to the corresponding customers by designing suitable catalogs for enterprises. Maximize the number of customers covered. In this paper, a new directory segmentation problem and an effective directory segmentation algorithm are proposed based on the in-depth study of the customer oriented directory segmentation problem. In the electronic commerce environment, the customer's purchase behavior is affected by many factors. Based on the online consumer's purchase psychology and behavior, this paper extends the problem of catalog segmentation with profit constraints. This paper proposes a dual-objective catalog segmentation problem considering freight, which aims at maximizing the expected profit and the number of covered customers. In addition, based on the customer value theory and considering the long-term benefit of the enterprise, a new perception-oriented catalog segmentation problem is proposed to improve the click-through rate and conversion rate of catalog commodities. In this paper, an improved cooperative genetic algorithm is proposed. The algorithm divides the population into two sub-populations to realize the evolution within the population and the coevolution between the populations, and designs the optimal solution set and the adaptive mutation operator. The search process of the algorithm can jump out of the local optimum and accelerate the convergence of the algorithm. In addition, the cooperative crossover operator is used to realize the information exchange between the two populations, which makes the two subpopulations adapt to each other and co-evolve. Based on the real e-commerce data, the proposed algorithm is compared with the memetic algorithm and the classical RBPF algorithm. The results show that the proposed algorithm can cover more customers and is better than the first two algorithms. Based on the proposed algorithm, this paper realizes the visualization of catalog customization through the prototype system.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F274
[Abstract]:In the Internet era, the amount of data increases explosively, so it is an urgent need to extract valuable information from massive data. In order to meet this demand, data mining technology emerges as the times require, and has attracted more and more scholars' attention. Under this background, many theories of data mining have been put forward, and the framework of data mining based on microeconomics is one of them. The problem of customer-oriented catalog segmentation is an important research topic of data mining based on microeconomic viewpoint in the business field. It is hoped that it can be sent to the corresponding customers by designing suitable catalogs for enterprises. Maximize the number of customers covered. In this paper, a new directory segmentation problem and an effective directory segmentation algorithm are proposed based on the in-depth study of the customer oriented directory segmentation problem. In the electronic commerce environment, the customer's purchase behavior is affected by many factors. Based on the online consumer's purchase psychology and behavior, this paper extends the problem of catalog segmentation with profit constraints. This paper proposes a dual-objective catalog segmentation problem considering freight, which aims at maximizing the expected profit and the number of covered customers. In addition, based on the customer value theory and considering the long-term benefit of the enterprise, a new perception-oriented catalog segmentation problem is proposed to improve the click-through rate and conversion rate of catalog commodities. In this paper, an improved cooperative genetic algorithm is proposed. The algorithm divides the population into two sub-populations to realize the evolution within the population and the coevolution between the populations, and designs the optimal solution set and the adaptive mutation operator. The search process of the algorithm can jump out of the local optimum and accelerate the convergence of the algorithm. In addition, the cooperative crossover operator is used to realize the information exchange between the two populations, which makes the two subpopulations adapt to each other and co-evolve. Based on the real e-commerce data, the proposed algorithm is compared with the memetic algorithm and the classical RBPF algorithm. The results show that the proposed algorithm can cover more customers and is better than the first two algorithms. Based on the proposed algorithm, this paper realizes the visualization of catalog customization through the prototype system.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F274
【参考文献】
相关期刊论文 前5条
1 张丽萍;陈玮;;面向寿险客户的目录分割问题研究[J];计算机工程;2010年10期
2 徐秀娟;王U，
本文编号：2250138
本文链接：https://www.wllwen.com/jingjilunwen/dianzishangwulunwen/2250138.html
