A Study On Credit Risk Assessment Model And The Implemention Of The Tool Based On Decision Tree

Credit Risk is the main factor to impact the safety and efficient operation of commercial banks. Business community, is regarded as a core component in Bank Loan business, its management, and codes of conduct is directly related to the utilization and return of the bank credit. So it requires banks to give a scientific evaluation on the activities, operations, profitability, solvency in business, to ensure if they should offer loans to the corporations. Currently, decision tree is the best and most popular in the ways to solve the classification problem.This article is an empirical research for the process of implementing a credit rating process tool based on the original decision tree, with a appropriate revision in it, and a demonstration to confirm the better efficiency and precision of the modified decision model compared with the traditional one.Firstly, the article finds out the shortcoming of the C4.5algorithm in how to deal with continuous attributes:In order to seek the best dividing point, C4.5algorithm need to scan and sort the data repeatedly, and its heavy computation makes the algorithm inefficient. Considering this, this article proposed a discretization algorithm based on boundary points’ attribute values merging and inconsistency checking. This method break the idea of traditional continual attribute discretization’s traversal heavy search, obviously enhanced the separate effect on the bases of efficiency. Secondly, this article also made a research on pruning algorithm of the decision tree and proposed a improved pruning algorithm based on PEP. Thirdly, the article made an empirical research to confirm the better efficiency and precision of the modified decision model compared with the traditional one. Finally, the article designed a simple tool for credit risk evaluation.