Analyze And Research Of E-commence User Behavior Based On Spark

With the rapid development of China's mobile Internet and the advent of the 5G era,employees from all walks of life will use e-commerce sites to purchase all kinds of goods they need in life.The log information generated by the click behavior of e-commerce users on the Internet is a huge gold mine.How to deal with and mine this gold mine has become an urgent problem to be solved and has attracted the attention of numerous scholars.In the face of massive user log information,the major problem facing researchers lies in data storage and data mining,and the emerging big data technology solves these two problems well.The whole big data ecosystem based on Hadoop platform contains data storage and data analysis functions.Hadoop's distributed file storage system HDFS is used to store massive user log information,and Spark's memory-based iterative computing model can quickly help us analyze and calculate user log information.In this paper,the Spark cluster is built to realize the functions of each module of the system by using relevant components of Spark ecosystem.SQL component implementation using Spark Core + Spark electric business user behavior data processing and statistics,use the Spark MLlib machine learning repository implementation parallelization of ALS(Alternating further Square)algorithm by the analysis of user behavior data mining.The offline computing and real-time computing of Spark cluster are combined to systematically analyze the behavior trajectory of Internet users by utilizing the powerful computing power of the cluster.Based on the data analyzed,a recommendation model combining offline recommendation and real-time recommendation is designed for users.