Design And Application Of Customer Precision Marketing Based On IFA Clustering Algorithm

With the application of Big Data and other Internet technologies in the financial industry,a wide variety of new financial products have a significant impact on the traditional banking business,especially Chinese banks,because the common people are used to applications such as Alipay,WeChat,which can also provide them financial products by simply clicking on the mobile phone.In the face of more and more fierce market competition,Chinese banks have changed their development mode from product-centered to customer-centered gradually.As the customers become the focus of market competition,the key to be successful in this competition has changed into how to understand the customers of the banks and provide the service which customers want.It is a mission which the clustering of customers can make it possible.First of all,if we want to do a successful clustering,we must select a series of indexes to describe the bank customers.In this thesis,a series of indexes,such as demographic characteristics,products held by the bank customers,transactions of the bank customers,are selected to achieve this goal.A detail description of calculation formula for each index is also provided in this thesis.On the basis of these definitions,the thesis creates a view of the bank customers which can reflect the characteristics of the customers.As some indexes are not suitable for the clustering,these indexes are screened through correlation analysis to exclude highly correlated indexes.In the second place,the thesis compares the performance of K-means algorithm,bisecting K-means algorithm,FA clustering algorithm,IFA clustering algorithm on the standard data set provided by UCI,in order to find the best clustering algorithm,and IFA clustering algorithm is the final choice in this thesis,because this algorithm has higher accuracy and is more stable.On the base of above-mentioned works,this thesis realizes the clustering of the bank customers.By analyzing indexes of each clustering group,the thesis summarizes the characteristics of each clustering group.According to all these characteristics concluded,different marketing plans and strategies for all these clustering groups are provided by the thesis.In order to make the analysis result play a more important role in the banking business,the existing customer relationship management system(CRM)is improved.First of all,the process of calculating all the indexes of bank customers at regular intervals is added for the view of the customers must be updated regularly to reflect the customers accurately.The module which users can carry out customer clustering work independently is also provided,and with its help,the users can cluster the customer by themselves when it is necessary.The system updated also provides the way which users can check the clustering result,and the module which users can maintain the clustering result,so the users can have a better understanding of the customers.A function with which the users can maintain the products to be marketed presently is also provided.If the maintenance of the clustering groups and the products is correct,the system can make matching the customers and the products to be marketed automatically.Combined with the improvement of the queuing system,the CRM updated can send the detail indexes about the customer who comes to the bank branch and his clustering result,and the users can also query the detail information about this customer on the system.With all these modules,the customer manager can provide a better service for the customer.Up till the present moment,the system runs well and does a big favor for the banking business.