Research On Commercial Bank's Customer Identifying And Customer Retention Models

Eeconomic Globalization and electronic commerce have made the competition more intensely which the commercial banks are confronted with. The ability to manage the customer assets efficiently has become to be a determinant factor that affects corporations'surviving and development. Therefore, the research on customer identifying and customer retention in commercial banks has significant value in theory and practice to recognize the nature of customer relationship management (CRM) and explore the operation mechanism of CRM. It is also important to improve the decision ability based on the fact and gain a true competitive advantage: the ability to manage the future and increase revenue by satisfying customer need in shorten response time and low cost and assigning resource to marketing campaign. Under the background, by using marketing, data mining and decision theory and methodology, the mechanism of customer relationship management, decision models and strategies and the decision methods in commercial banks were investigated systematically as following. (1) The analysis of customer value management in commercial banks The problems that commercial banks are faced with were described firstly. Pointing out that the essential of customer relationship management is customer value management, the customer value in commercial banks was analyzed and the frame of customer value management was put forward secondly. Finally, the support technologies of customer identifying and customer retention researches were set forward. (2) Customer identifying models of commercial banks based on segmentation-level. At first, this dissertation analyzed which factors were appropriate to be used in market segmentation by commercial banks and classified the factors by the extent to which each factors (variable) might be deemed observable, then described the characteristic of each kind of factors. Secondly, the uses of neural networks and clustering algorithm as the means of segmenting and identifying customers were examined on the segmentation level. In particular, the model of customer identifying based on expected benefits and attitudes in banks was built by using an agglomerative hierarchical method of merging. Meanwhile, the process to develop profitable customer segmentations in commercial banks was described with data mining technology based on the large database and the characteristics of each customer clusters were analyzed through numerical example. (3) The customer identifying models based on customer value from individual-level. The uses of credit scoring models and behavioral scoring models as the means of identifying customers based on individual-level were investigated firstly. By using three dimensions, current value, potential value, and customer loyalty, a customer lifetime value model of individual customer was established secondly. A customer identifying model based on customer value was built from individual-level. Thirdly, the methods by which profitable customers could be identified and segmented were explored through numerical example and corresponding marketing strategies were established by using the customer identifying result. (4) The model of identifying the fraud claim By suppose that previous claims may have been misclassified, but only in one direction, only honest claims may contain a portion of fraudulent claims that cannot be identified, the feasibility of using binary response models to identify the fraud claim was explored and the standard binary response model was expanded by increasing the omission errors parameter firstly. By doing so, this new model could account for the shortcomings of the original classification system and improve the performance of the model. Secondly, this modified model was used to identify the fraud claims based on a real estate mortgage database from a commercial bank that contained honest and fraudulent claims and the validity of the proposed model was demonstrated by comparingthe estimation results for the logit model with and without omission error. (5) The customer retention models The use of logistic regress analysis as a mean of building the customer retention models was explored firstly. Secondly, by examining the attributes that affected customer attrition used in other research and applying cross-tabulation technology to identify the attributes and variables that affected significantly customer defection, a model of customer defection in the commercial bank was developed with logistic regress analysis technology and the predictive effectiveness were compared by using the same population but under different misclassification costs. Meanwhile the characteristics of customers at risk of leaving in commercial bank were identified by using the data currently stored in the bank's database. Finally, a conclusion was draw and some issues were pointed out for future research in the field of customer identifying and customer retention.