基于Hadoop的数据挖掘在电商环境的研究与应用

发布时间：2018-03-25 02:39

本文选题：数据挖掘　切入点：关联规则算法　出处：《湖南大学》2016年硕士论文

【摘要】：随着便携式网络接入设备的飞速发展以及互联网技术的迭代更新,使得网络生态系统逐渐壮大、活跃,这也使得依托于互联网技术的电子商务发展迅速。相较于传统线下的购物方式,线上电子商务无疑是一种快捷、高效和便利的购物方式。近年来井喷的电商购物平台也很好的印证了这一点。对于电子商务平台的运营者来说,如何巩固现有客户、拓展潜在客户是重中之重。基于互联网时代快速、海量数据的特点,本文设计将数据挖掘技术应用于电商平台数据,一方面,深度发掘现有客户的浏览、购物习惯,巩固现有用户;另一方面,分析潜在用户行为,获取其兴趣点,进行定向推送,拓展更多的客户。基于电商平台用户购物数据之间存在较强的关联性,本文设计采用关联规则算法进行数据挖掘与分析,达到巩固现有用户,发掘新用户的目的。数据挖掘的过程就是发现隐藏在各种尚没有处理的原始数据集合中的各种相关联系,并从这些联系中提取知识的过程。数据挖掘是多种计算机相关学科相结合的产物,其包含了数据库技术、计算机机器自主学习、数据统计分析、行为模式识别、人工神经网络等等学科。由于其具有很高的商业使用价值,同时适合应用的范围极为广泛,所以目前数据挖掘的相关研究已成为研究的重点之一。本文以现今互联网、大数据时代下的电商平台为切入点,对电商平台现状进行分析,得出其弊端,即无法应对大数据时代海量无序数据的冲击,容易使平台积累无效数据,造成资源使用率低下,平台电商有效转化率低。其次,作者对某知名电商平台的服饰卖家以及家电卖家进行了匿名访谈,得出了服装买家购买物品具有较高关联度的结论。技术上,本文基于数据挖掘技术提出了一套基于Aprior i的关联规则算法,并利用Hadoop数据库集群进行数据处理,相较于传统的关系型数据库,Hadoop集群能同时对数据进行处理,大大提高算法工作效率。本文还基于Angular JS、Bootstrap以及Html搭建了一套前端数据可视化系统。
[Abstract]:With the rapid development of portable network access devices and the iterative updating of Internet technology, the network ecosystem is gradually expanding and active. This also makes e-commerce based on Internet technology develop rapidly. Compared with traditional offline shopping, online e-commerce is undoubtedly a kind of fast. Efficient and convenient shopping methods. In recent years, the blowout e-commerce shopping platform is also very good proof of this. For e-commerce platform operators, how to consolidate existing customers, Expanding potential customers is the most important thing. Based on the characteristics of fast and massive data in the Internet era, this paper designs and applies data mining technology to e-commerce platform data. On the one hand, it deeply excavates the browsing and shopping habits of existing customers. Consolidation of existing users; on the other hand, analysis of potential user behavior, access to their points of interest, directed push, expand the number of customers. Based on e-commerce platform, there is a strong correlation between user shopping data, In this paper, the association rule algorithm is used for data mining and analysis to consolidate existing users and discover new users. The process of data mining is to discover all kinds of related connections hidden in all kinds of raw data sets that have not yet been processed. The process of extracting knowledge from these links. Data mining is a combination of many computer related disciplines, including database technology, computer machine autonomous learning, data statistical analysis, behavior pattern recognition, Artificial neural network and other disciplines. Because of its high commercial value, and suitable for a wide range of applications, the current data mining related research has become one of the focus of research. Based on the analysis of the current situation of the e-commerce platform in big data's time, the author finds out its disadvantages, that is, it can not cope with the impact of the massive disordered data in the era of big data, which easily makes the platform accumulate invalid data, resulting in the low utilization rate of resources. Secondly, the author conducted anonymous interviews with clothing sellers and home appliance sellers of a well-known e-commerce platform, and drew the conclusion that clothing buyers have a high degree of correlation. In this paper, a set of association rules algorithm based on Aprior I is proposed based on data mining technology, and the data is processed by using Hadoop database cluster. Compared with the traditional relational database cluster, it can process the data at the same time. This paper also builds a front-end data visualization system based on Angular JS bootstrap and Html.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F724.6;TP311.13

【参考文献】