Separation and ranking of Refah bank’s good credit customers by the procedure data mining

Today data mining has been known as the most crucial technology for efficient and accurate exploitation of massive data and its importance is increasing day after day. One of the data mining’s functions is making some models for prediction of an objects class , according to some of its features. This term may mean a customer , transaction , family , message of electronic post and …. . For loan applicants , an object can also be divided into some classes , such as good or bad credit. The problem is , there are some customers , that despite their good credit , they are behaved like the other usual customers. For granting convenience , numerous gurantees are received from them and even , sometimes good credit customers , in comparison to bad (not very good credit) customers , receive their benedictory convenience sooner. In this paper , by the separation of good credit customers from not very good ones and their ranking , we aim to organize the way of bank’s behavior with customers (according to each customers rank) in terms of received gurantee amount , time priority of convenience and loan’s payment. In this paper , customer’s credit scoring will be proceeded with the consideration of loan applicants features and application of data mining. By the usage of the most efficient techniques of data mining (in case of classification) such as methods of decision tree (c5 , CART) and neural network and also with the help of clementine software , the data will be analyzed and the final model will be chosen by the comparison of three methods. Consequently , customers ranking prediction will be done and the way of bank’s behavior with customers will be assigned by the application of decision tree.

کلیدواژه ها:

data mining ، classification ، customers credit scoring ، decision tree ، neural network