The Application Of Credit Evaluating Model In Individual Residential Mortgage Based On Decision Tree

In1998, the State Council issued regulation aimed at deepening reform in urbanresidential markets and accelerating the construction of residential housing. Since then, Chinese mortgage markets had experienced a dramatic growth becoming a major element of lending. By the end of2012, commercial bank holdings of individual residential mortgages reached7.5trillion yuan.However as economic growth moderates andthe nonperforming-loan ratio (NPL) rises, the banking sector’s exposure represents agrowing risk, particularly to rural banks and credit markets. New measures are needed to manage the credit risk represented by individual residential mortgage loans. Thispaper proposes a potential solution using decision tree to model credit risk of mortgage contracts.This paper is divided into four sections. I first provide background on theidiosyncrasies and risks of the mortgage markets based on the algorithmic andstatistical principles of the decision tree model. In the following section this paperanalyzes the credit risk rating mechanism and builds the credit risk rating tablethrough the introduction of Relative Importance Degree and Credit Risk StandardMeasurement. In the third section, I explain how to apply the decision tree to create acredit evaluating model. I used SPSS Clementine12.0implement the model andconstruct credit risk rating tables and predictions. In the last section I evaluate themodel through empirical analysis of historical residential mortgage data. Through themethods using simple statistical methods and Logistic regression I get the results。Theresults validated the decision tree model, and verified the prediction accuracy of themodel.