Research And Implementation Of Bank Personalized Service Mode Based On Data Mining

With the development of both Internet technology and globalization,the challenges faced by bank in many different aspects are as follows:Firstly the market competition has become much fiercer and at the same time the difference of products among different banks is not apparent,so the advantages based on products is not apparent as before.Secondly customer's requirement for personalization and diversification has become much stronger.The simplification of the product and service of business bank will not meet customer's requirement.The last but not least is that the communication in Internet has become much simpler and feasible.All kinds of products and service can show their information in terms of quality and prices as well as market,which provide a lot of choice for customers.So the modern banks want to remain competitive in the highly competitive market,they must use information and intelligent method,so they can understand what the customers need,evaluate the customers value,mining the people who are potential customers and develop key customers.In addition,they also can provide high quality,personalized products and services to customers,improving their customers' satisfaction and allegiance.So the personalized service based on data mining is the route bank must take.This article analyze customer's basic information and customer operation log and other valuable information on the customer segmentation using the fuzzy clustering technology and Bayesian network techniques in data mining,and establish user subdivision model which can distinguish the potential value of the different partitioned segments.Finally help related personnel of banks to implement personalized service,improve the bank's core competitiveness through the customer segmentation model.This paper use the bank customers' record as the research object,7 most valuable information attributes are selected as the attribute set of training model through the analysis of the basic information of customer records and operation log information.And the segmentation model is established based on above two kinds of data mining methods and at the same time we give the evaluation of segmentation results.The main research content has two parts.In part one,we use the Fuzzy cluster method and Bayes network to build the Customer Segmentation model,in order to objectively and scientifically classify customers on the basis of their characteristics.Firstly,Cluster the customer's samples and divide the samples into several disparate groups,then train Bayes network in each groups,finally establish the effective segmentation model.In part two,we Research the feasible customer recommended optimization model,which can satisfy customers personalized service requirements.For a personalized service we use the customer segmentation model,the cost parameters and the earnings parameters to evaluate whether it is feasible.Finally,we randomly selected 5000 sample data from a database,then 2000 out of the data which are cleaned and preprocessed are selected as the training model of the sample data set U and at the same time,we selected 1000 samples as a test sample data set T,of which the number of wealth management products of A is 400,the number of B 300 financial products,the number of 300 fund products C.We find that when ?=0.75,the overall classification accuracy of the models is the highest,and the results of the experiment also shows that the customers personalized model is effective and can be used to provide scientific decision support for bank policymakers.