Credit Risk Assessment Model Based on Domain Knowledge Constraint

Abstract

With the continuous rising of real-estate prices and the upsurge demands by residents, the loan default risk has been raised gradually due to the individual housing loan increased with years. The efficient measurement and management systems for the credit risk in individual loan should be urgently established. Such systems need a knowledge-based decision methodology to be implemented. The decision tree algorithm is one of methods. It is applicable to enhance the riskpsilas assessment of Chinese individual real-estate loan. It has several advantages such as understandable principle, low demand, and interpretable results that can be visualized. In this paper, the decision tree and information entropy theories are applied to the credit-risk assessment of individual housing. Based on the theory of decision tree and domain knowledge, the evaluation of attribute to measure important degrees by knowledge-based information gained and a theoretical structure equation was established. It was found that using such approach, a higher accuracy for forecasts can be reached.