针对网购用户行为的分析与研究

发布时间：2018-05-04 02:14

本文选题：用户浏览行为 + 数据挖掘　；参考：《西安理工大学》2016年硕士论文

【摘要】：随着我国科技的发展和人们生活节奏的加快,网络购物以其在时间、地域及商品的多样选择等方面占有的优势而受到众多网民的青睐。我国的网络购物市场规模的扩大和网上购物者数量的激增促使电子商务平台产生大量的电子商务浏览和销售数据。面对纷繁复杂的浏览和销售数据,商家需要进行深入的分析和理解数据并从中挖掘出用户潜在的需求,从而制定出各种销售策略以满足用户的需求,进而为电子商务平台赢得更多的顾客。为此,本文将从电子商务用户的浏览和购买行为数据着手,利用统计分析和数据挖掘的方法对用户的行为数据进行分析和研究。已有的研究表明,用户的在线浏览轨迹与在线购买行为之间存在着较强的相关联特性。因此通过数据挖掘的方法对搜集到的用户行为轨迹数据进行分析和研究,可以从中挖掘出用户倾向购买的商品,即是用户感兴趣的商品,从而可以很好的了解用户的需求。本文主要工作如下:(1)本文将用户的购买行为归结为一种用户浏览行为累积之后的行为。本文首先对搜集到的大量用户行为轨迹数据进行统计分析,结合分析结果采用一种基于商品信息素的挖掘算法,该算法思想来自于蚁群算法中信息素的思想,这里商品信息素的值代表用户对商品的感兴趣程度。(2)本文在以往研究的基础上对商品信息素的求解算法进行了改进,因为以往在求解商品信息素时没有考虑到商品所对应的浏览、收藏、放入购物车、购买这几种不同的用户行为,用户行为的不同代表着用户对商品感兴趣程度的不同,所以在求商品信息素时不能视为一样,应区别对待。又由于兴趣会随着时间的增加而逐渐衰减,在信息素的求解上加入了衰减因子,使得一些已经不流行的商品不至于一直拥有很高的商品信息素,形成良好的新旧交替的情况。(3)最后是实验部分,通过实验来对比本文改进的方法在用户行为数据分析上的优势,结果显示改进后的实验效果明显好于未改进之前的实验效果,同时又使得数据得到最大化利用。
[Abstract]:With the development of science and technology in our country and the quickening of people's life rhythm, online shopping is favored by many netizens because of its advantages in time, region and various choice of commodities. The expansion of China's online shopping market and the rapid increase in the number of online shoppers make e-commerce platform produce a large number of e-commerce browsing and sales data. In the face of complicated browsing and sales data, merchants need to deeply analyze and understand the data and dig out the potential needs of users, so as to formulate a variety of sales strategies to meet the needs of users. And then win more customers for the e-commerce platform. Therefore, this paper starts with the browsing and purchasing behavior data of e-commerce users, and makes use of the methods of statistical analysis and data mining to analyze and study the behavior data of users. Previous studies have shown that there is a strong correlation between online browsing trajectory and online purchasing behavior. Therefore through the data mining method to collect the user behavior track data to carry on the analysis and the research may excavate the product which the user tends to buy namely is the user is interested in the commodity thus can understand the user's demand very well. The main work of this paper is as follows: (1) in this paper, the user's purchase behavior is reduced to a behavior after the user's browsing behavior accumulates. In this paper, a large number of user behavior trajectory data collected from the statistical analysis, combined with the results of the analysis using a commodity pheromone mining algorithm, the idea of the algorithm is derived from the ant colony algorithm pheromone idea. Here, the value of commodity pheromone represents the degree of interest of the user to the commodity.) in this paper, we improve the algorithm of solving commodity pheromone based on the previous research. Because in the past, when calculating commodity pheromones, they did not take into account the corresponding browsing, collecting, putting in shopping carts, buying these different user behaviors, the different user behavior represents the different degree of interest of the user in the goods. So in seeking commodity pheromone can not be regarded as the same, should be treated differently. And because interest gradually attenuates over time, a decay factor is added to the solution of pheromones, so that some commodities that are no longer popular do not always have very high commodity pheromones. Finally, the experiment part is to compare the advantages of the improved method in the analysis of user behavior data. The results show that the experimental effect of the improved method is obviously better than that of the experiment before the improvement. At the same time, it maximizes the use of data.
【学位授予单位】：西安理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F724.6;F713.55;TP311.13

【参考文献】