Applied Research Of Data Mining In Customer Relationship Management

Along with the change of traditional principle of management,the center of gravity of business enterprise has transferred from "product" to "customer",and the resource of customers has become the most important property of business enterprises, and CRM becomes the important tool which helps business enterprises to understand customers and grasp the market.Now most of CRM used in our country is operational CRM,used for informationization and integration of the each section business process in the business enterprises.Along with carrying operational CRM usage,the business enterprises will accumulate more and more customers' information and the integration, excavation and knowledge discovery in such information become a kind of inevitable need.Data Mining is a kind of new processing technique of the business information and its main characteristic is extracting,converting,analysizing and processing with different models on a great deal of business data in the business database and withdrawing key data that support for business' decision.The modern enterprise's competition is based on the service and the information competition,Data Mining technology can effectively excavate the useful information and the knowledge from the massive customer data,then can effectively promote the quality of CRM,at last can achieve the goal of enhancing the enterprises competitive power.This paper sets out from the CRM basic principle and studies the structure of Customer Relationship Management and constitution and data mining application in it. On this foundation,a method based on the Decision Tree algorithm to create customer run off analysis model was proposed,and an improvement to classic C4.5 algorithm was carried on.Before setting up the model,it is necessary to carry on the clean,the transformation and the revision to the data.And also only like this,it can be real to display the power of algorithm.In view of flaw values,according to the field importance and the flaw values situation,this paper uses many kinds of processing mode to make the data more reasonable and the mining quality improved.