Credit scoring model: a comparative study based on Chinese real credit data

摘要

Credit scoring is one of the important credit risk management techniques on consumer credit. The result of credit scoring model is more influenced by the economy,cultural,society,etc.,so filtering the better credit scoring model is key. This article reviews the former credit scoring techniques and studies comparatively the three models of logistic,artificial neural network and classification tree based on the real credit data from a Chinese city. It points out that there are different influence factors to credit risk in different region,which decides the importance of filtering model. The result shows also that classification tree model is the best in today's Chinese environment,but the result is used carefully in other region according to above factors.