The Design And Implementation Of Bank Customer Rating System Based On Data Mining

In recent years,with the vigorous economic development,the demand for customer loans in the banking industry is increasing year by year.Under this circumstance,it is important for commercial banks to do a good job on customer risk control and improve the establishment of customer rating system.Traditional methods of customer risk control and customer rating are not effective with the rapid development of Internet financial environment.This paper designs and implements a bank customer rating system based on decision tree algorithm in data mining.The main work includes:Firstly,the current customer rating workflow is systematically analyzed and investigated,and the difficulties encountered in the current customer rating process are collected and sorted out.In view of the problems in the current rating process,such as the decentralization of rating data sources,the greater influence of subjective factors of rating personnel,and the large amount of supplementary work,the corresponding solutions are put forward.Secondly,Pentaho is used as ETL tool to extract data from each system,load the data needed for rating into customer rating system,and integrate multi-system rating data.The CART decision tree algorithm is used to train and generate customer rating model.The model is used to classify bank customers,avoiding the subjective influence of raters on customer rating results.It improves the mechanism of supplement and the storage mode of supplement data,avoids repetitive supplement,reduces the supplement content and the supplement workload of relevant personnel,and shortens the overall working time of rating.In addition,there are two innovations in the way of updating training data sets:as the data environment changes constantly,the manual adjustment workload of graders will be increased when the classification accuracy is reduced.This system will update the manual adjustment data into training data;the training data set will be updated dynamically according to the data collection time,so that the outdated data will be eliminated and the timeliness of training data will be improved.Finally,the system adopts Spring MVC architecture,uses Spring Boot to build the system,and uses Oracle,a relational database,to build the database cluster.The main functions of the system,such as customer information management,customer rating management,data processing,data interface and so on,are designed and implemented.This paper applies data mining technology to the process of bank customer rating,makes the results of bank customer rating more objective and scientific,and makes the operation of rating more efficient and convenient.