Development And Technical Research Of Scoring Model

With the prosperity and development of the commodity economy and the continuous deepening of the market economy,personal credit business has become the key area of the development of commercial banks in China.How to find the valuable rules from the complex credit customer data has become a major challenge for the credit risk management of commercial banks.The use of data mining technology to establish a personal credit scoring model,can accurately predict customer default probability,distinguish customer "good" "bad",so as to improve the competitiveness of commercial banks products and services,ultimately optimize the whole society's resource allocation,stimulate the country's consumption growth.This paper studies the basic theory of data mining and credit scoring,and carries out an empirical analysis on the original customer data of a commercial bank(80000 observations,627 variables).Using SAS to preprocess data,build model using EM,use decision tree,logical regression and neural network to establish individual credit score combination model and single model respectively,and make a comparison and analysis of the model.