Research And Development Of Online Retailers Customer Data Analysis System

For online retailers,they mainly focus on accurate customer-classification management and precise operation.However,up to now,this kind of customer-classification management in e-commerce operation is seriously relying on experiences from operating teams.This leads to vagueness in customer classification and lack of credibility.The final results are massive loss of customers,decreasing of customer stickiness as well as sharp declining in revenue for online retailers.Nowadays,as technology develops,the retailers make efforts to collect,process and analyze data of all users in order to realize accurate operation.But the problem is a large number of data falls into different servers,which results in difficult data intercommunication,inaccurate analysis and rough customer-classification system.Given this background,it is urgent to bring in data-analyzing system on customers.It lays a good technical foundation for data management,accurate operation and efficient data analysis.At present,with the development of technologies,data-analyzing system on customers has been widely used.It bases itself largely on three parts,namely how to build a highly efficient system,provide better services for online retailers and effectively classify the customers.The paper,founding on key relative technologies,will divide e-commerce data-analyzing system into three modules for discussion.They include modules for data acquisition,centralization,and analysis.In terms of acquisition module,it takes Filebeat and Sqoop respectively as log-collection agent and data-synchronization method;the second module mainly focuses on Kafka and HDFS;the analysis part uses Hive and HBase as database.It takes advantages of Spark to perform mathematical operation,applying Java web as a way of showing.Therefore,the paper,with the help of above technically-building system,uses RFM model and K-means to classify customers of online retailers.This system slove the data collecting,data storge and data analysis problem.By using filebeat,log collecting is more efficiency.And System become more and more effencieny which using the Spark as calculation platform.So that,the system has certain practical and promotional value.