Research And Application Of Data-mining Based On Hadoop In E-commerce

With the development of portable net access device and the revolution of the Internet technology,the net echo-system becomes active,which is the base of the fast development of E-commerce.Compared with the traditional off-line shopping mode,the E-commerce is a fast and convenient way.As an executor of a E-commercial platform,how to confirm our clients and to find more potential clients is the key.Based on the mass data of the Internet,the thesis tries to use the data mining on a E-commercial platform.The first goal is to find out the browsing patterns and purchasing behaviors,which could maintain old users by more suitable service.The other goal is to analyze the potential users' behavior and interests,which could get more new users by accurate marketing.The thesis applies association rules mining algorithm to maintain old users and to get more new users.The process of data mining is to find valued knowledge behind the unmined data.Data mining is a cross-subject technology contains database,machine learning,statistics,pattern recognition and artificial neural network.Because of its commercial value,data mining has become an important research.The thesisanalyzes the status of the whole E-commercial,and find out many of them could not execute so many data effectively.This may accumulatetoo much useless data and cause the inefficiency and low conversion rate.Then the author has an anonymous interview with one famous E-commercial platform store owner,and gets the high-association rules of consumption behaviors.Technically,the thesis codes a set of Apriori algorithm bases on Hadoop dataset.Compared with the relational database,Hadoop can execute on multi-servers simultaneously,which can improve the algorithm efficiency.What's more,the thesis builds a data visualization system by applying AngularJS,Bootstrap and Html5.