Application Research On Data Mining In Customer Relationship Management Of Commercial Bank

Customer relationship management plays an important role in banking management,as akey technology,data mining plays a decisive role in analysis and implementation of the bankcustomer relationship management. At present to bank data mining techniques in customerrelationship management research mainly includes the data mining algorithm and process.The algorithm can be used to extract the potential information in the database,but theaccuracy and effectiveness remains to be improved. In the same way, implement process fordata mining research is confined to the traditional two models.This article will improve onthese two aspects, put forward the improved algorithm and implementation, and applied tothe customer relationship management.In view of the data mining algorithm, this paper selected the clustering analysis methodas a representative,put forward a improved algorithm in combination with hierarchicalclustering and K-means clustering method.On the one hand, both solved the class objectcannot exchange and poor scalability in hierarchical clustering method.On the otherhand,solved the K-means clustering method is sensitive to the initial clustering center.At thesame time,using the test data of UCI database to test also shows that the improved clusteringanalysis algorithm clustering effect is better.According to the implementation of the data mining process, this paper develop A newapproach combining the SEMMA method with GRISP-DM process model.Exploitingrespective advantage to both the GRISP-DM method can comprehensively summarized abusiness,and the SEMMA method can accurately analysis the data.In addition, it is proved tobe more scientific that the improved model applied in customer churn warning. Customer segmentation is an important part of the bank customer relationshipmanagement,this paper study on the credit card customer information with the improvedalgorithm and process.At last,five kinds of credit card customers are distinguished andtargeted solutions have been proposed.