Visual Analysis of E-Commerce User Behavior Based on Log Mining

Abstract

With the continuous development of internet economy and e-commerce, the scale of data produced by users on e-commerce platform is increasing explosively. Mining the behavior of individual users and group users from massive user behavior data and analyzing the value and law behind the data are of great significance to the development of e-commerce. Taking the user behavior log data of an e-commerce website as the data source, this paper, firstly, processes and analyzes the original dataset through the data filtering and storage module, and it uses the combination of Kafka and Flume to store the user behavior log with reasonable structure and complete fields in HDFS. Secondly, a hierarchical system of data warehouse is constructed in Hive, and each layer of log data is effectively mined and multidimensionally analyzed with the help of log mining technology. Finally, based on the big data framework and Bi tools, a data warehouse system is designed and implemented, which could store and analyze massive data and visually display the results. The system uses dimensional modeling to build a data warehouse hierarchical system to mine and analyze user behavior data through log mining algorithm deeply. The K-means clustering algorithm and RFM model are used to divide the user behavior characteristics in detail, and AARRR funnel model is used to analyze the logs in a modular way. Through the effective mining and multidimensional visual analysis of user behavior data, the behavior analysis of group users and individual users, as well as the analysis of commodity sales flow and sales linkage are realized, which provides support for internal decision-making and precision marketing.