Applied Research On Data Mining In Bank Customer Churn

In bank product homogeneity phenomenon universal existence today, the way of customer choose product and service becomes various and the customer loyalty is more and lower. The customer churn already becomes one of the most matters of concern in banking industry. Through analyzing and forecasting the characteristic of customer churn behavior and carrying on the effective management, the reducing of customer churn as far as possible is an urgent question.This paper uses the method which the fundamental research and the real diagnosis research unifies, based on question and so on question as well as domestic and foreign applications condition which to the customer relations management present situation, faces discussion, how application data mining technology in detail elaborated to drain the question to the customer to carry on the modeling and the forecast. The paper obtains from the data mining and the CRM elementary theory, take the banking industry as the background, take the Chinese construction bank some branch's customer transaction data as the foundation, utilizes the LOGISTIC return model, described one kind to excavate under the platform with emphasis based on SAS9.1 the customer to drain the modeling process. In detail narrated the customer to drain the model the structure mentality, drained the commercial limits from the customer, needed the data set, the modeling variable choice to the modeling, when essential factor the and so on compartment determination, carried on drains the model the system design work. In the actual excavation process, to include the data the preparation, the clean, the variable establishment, the model constructs as well as the model appraisal entire modeling process has carried on the multi-analysis and the research.The paper from the preliminary screening 30 forecasts variable, carries on the return operation determination after the SAS9.1 programming, in the final conclusion determined affects 5 contributions biggest forecasts variable which the bank customerdrains, and drains the model from this the establishment. In view of each customer, after leads the customer actual transaction data the model, may obtain each customer to drain the probability and to drain the scale division, and from this carries on has the pointed marketing, then for the bank management, the decision-making provides the reliable quantification basis.Drains the model through the establishment customer, forecast the key aspect which customers most have which drains the tendency as well as the influence customer drain, helps the recognition risk customer before, drains in the customer takes the pointed measure to detain them. This will change the bank formerly to obtain the customer in the success later to be unable to monitor the customer to drain, to be unable effectively to realize the customer concern condition. How finally discussed has carried on the decision-making using the model assistance enterprise superintendents, and discussed the customer from the customer relations marketing angle to maintain the strategy and to propose the reduced customer drained the concrete measure, proposed waits for direction or topic which further studied. The paper applies the data mining technology to the bank customer churn well in the forecast, has obtained the good effect.