A Model Based on Combination of Neural Network and Decision Tree for Credit Risk Assessment

The paper builds a credit risk assessment model based on combination of neural network and decision tree. The method ranks financial data of loan enterprise based on the importance of the attributes, prunes the attributes using RBF neural network and builds a decision tree. Therefore conclusions of classification can be drawn as to whether or not enterprises break a contract. Finally comparing NN-DT with methods of discriminate analysis and C4.5 algorithm, studies are empirically carried out, and NN-DT largely improves the prediction precision.