| In recent years,with the improvement of customer requirements for personalized consumption,in order to obtain competitive advantage,enterprises put the personalized needs of customers in the very important position.The personalized needs of customers cannot be achieved without the analysis of customer behavior clustering data,but the customer data type is complicated,with huge number,when making personalized clustering mining,there are multiple difficulties for choosing data mining methods and data mining platform.So,the solution of choosing right customer behavior data mining method and platform is an urgency matter.Therefore,in this paper,the Xyz electric business insurance platform is taken as the breakthrough point-to solve the problem of clustering algorithm in mining and platform of Xyz platform,providing the basis for personalized service to customers,so as to gain a competitive advantage in the market.In order to solve Xyz platform customer behavior problems in the process of data mining,this article first analyzes the current research situation of big data,data mining,clustering,and customer behavior,and summarizes the current common research methods,lays a solid theoretical basis.Then,in order to research further on the background of the Xyz customer behavior,the paper analyzes the frame structure,the shopping process,the process of the service,customer behavior,and customer characteristics of the platform,which is important for Xyz customer clustering mining.On the basis of summarizing and analyzing the existing clustering algorithms,this paper bases on the characteristics of Xyz customer data,takes K-means as the basis of clustering analysis algorithm.By improving initialization,selection strategy,adjustment strategy and the new generation mechanism of artificial colony algorithm,the paper improves the shortage of original artificial colony algorithm,and then combines the improved algorithm with K-means algorithm for K-ABC algorithm,so as to compensate for the defects of K-means algorithm which is depending on the initial value,to enhance the reliability and effectiveness of the algorithm.According to the characteristics of the selection algorithm and the comparison of the current data processing platform,the paper selects the Hadoop as the data processing platform,and deploys the clustering method of parallel.Finally,this paper conducts an experiment on the Hadoop platform with Xyz customer data,proves the effectiveness of clustering algorithm K-ABC,and implements successful Xyz customer clustering.This article improves initialization,selection strategy,adjustment strategy and the new generation mechanism of the classical artificial colony algorithm,improves the efficiency and robustness of artificial colony algorithm.At the same time,the article combines improved artificial colony algorithm with K-means algorithm,makes up the defect of K-means algorithm.The research of the paper provides a certain technical support for many companies to conduct customer clustering mining with customer behavior data. |
