基于证券理财产品用户行为分析的个性化推荐研究

发布时间：2018-01-12 11:39

本文关键词：基于证券理财产品用户行为分析的个性化推荐研究　出处：《电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：在大数据的时代背景下,零售、金融和医疗等传统行业开始向“数据驱动型企业”转型,他们意识到,对用户在线行为的记录以及对其意图和偏好的挖掘可以为企业的运营和营销提供强有力的支持。证券行业自身的业务形态产生了大量质量高、价值大的数据,具有极大的挖掘价值。随着人们理财意识的加强,对于用户理财行为特征的研究也逐渐受到更多关注。随着科技的迅猛发展,证券用户可以获取的信息量爆炸性的增长,但我们选择信息的能力有限,如何通过个性化推荐技术解决证券用户遭遇的信息过载现象,成为了证券公司和理财用户急需解决的问题。本文既是在此背景下,通过分析证券理财产品用户的行为模式,研究适用于理财产品的个性化推荐方式,进而形成一个完整的证券理财个性化推荐系统。本文的主要工作可以概括为以下三点:(1)分析了用户购买理财产品的数据,统计了用户购买次数、理财产品销售情况和用户的活跃性,并从人类动力学的角度挖掘了用户购买理财产品的行为特征,同其他人类行为一样,主要表现为“强阵发弱记忆”的特性。(2)研究了常用的协同过滤和混合扩散算法,并将这些算法应用于证券理财产品的个性化推荐中,同时根据证券理财用户的使用场景,提出了基于用户聚类的热门推荐和基于用户实时行为的个性化推荐两种扩展的推荐策略。(3)参与设计并实现了基于证券理财产品的个性化推荐系统,详细说明了个性化理财产品营销子模块的框架设计,以及离线分析和在线推荐的处理流程,通过增量推荐的方式达到秒级内的响应时间,并从算法和营销两个角度,实现了个性化推荐算法在推荐效果上的提升。
[Abstract]:In the context of the era of big data, traditional industries such as retail, finance and healthcare began to shift to "data-driven enterprises," and they realized that. The recording of users' online behavior and the mining of their intentions and preferences can provide strong support for the operation and marketing of enterprises. The business form of the securities industry produces a large amount of high-quality and valuable data. With the strengthening of people's awareness of financial management, the research on the characteristics of user's financial management behavior has gradually received more attention. With the rapid development of science and technology. The amount of information that securities users can obtain increases explosively, but our ability to choose information is limited, how to solve the information overload phenomenon encountered by securities users through personalized recommendation technology. This paper is based on the analysis of the behavior patterns of the users of securities financial products, and studies the individualized recommendation methods suitable for the financial products. The main work of this paper can be summarized as the following three points: 1) analyze the data of the purchase of financial products by users, and count the number of purchases by users. The sales of financial products and the activity of users, and from the perspective of human dynamics to explore the behavior of users to buy financial products, just like other human behavior. The characteristics of "strong burst weak memory". (2) the common collaborative filtering and mixed diffusion algorithms are studied, and these algorithms are applied to the personalized recommendation of securities financing products. At the same time, according to the use of securities financial users. In this paper, two extended recommendation strategies based on user clustering and personalized recommendation based on real-time behavior are proposed to design and implement the personalized recommendation system based on securities financial products. The frame design of marketing sub-module of personalized financial management product and the processing flow of offline analysis and online recommendation are described in detail. The response time within seconds is achieved by incremental recommendation. And from the two angles of algorithm and marketing, the personalized recommendation algorithm in the promotion of recommendation effect.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.3;F830.9

【参考文献】