Research On The Index And Model Of Personal Credit Scoring

With the rapid development of our country's economy,people's consumption concept has also undergone tremendous changes.Housing mortgage,credit cards,car loans and other personal consumer loans' demand is continuously getting higher,which is an important profit growth point for commercial banks.In this context,commercial banks need to promote the development of personal credit business and innovation actively,and personal credit assessment is the key to the development of personal credit business.At present,China's personal credit system is still in its infancy,various commercial banks have not yet formed a unified personal credit evaluation index system and did not make a comprehensive consideration in the assessment of the choice of indicators.China's commercial banks often overlooked the applicant's credit history,personal tax records,judicial records and other lost record to take into account in the credit audit process,which can not carry out a complete credit evaluation and will increased the credit risk of commercial banks virtually.In addition,the current level of credit risk management of commercial banks in China is insufficient,and the efficiency and accuracy of personal credit assessment have yet to be improved,therefor we need a more scientific and effective method of personal credit evaluation.In order to reduce the credit risk of banks,in the process of carrying out personal credit business,we need to build a comprehensive and comprehensive evaluation index system of personal credit,and use the scientific and efficient personal credit evaluation method to evaluate the credit status of loan applicants objectively.In view of this,this paper starts from two aspects of personal credit evaluation index and model.First of all,on the basis of the related theories,this paper summarizes the five principles of constructing the personal credit evaluation index system.According to the personal credit evaluation standards both at home and abroad,this article compares the current personal credit evaluation index of commercial banks in China with the developed countries and P2 P lending platforms.The article builds a personal credit evaluation index system composed of five first-level indicators including natural conditions,family conditions,occupations,bank relations,credit conditions and 24 second-level indicators.The index system involves comprehensive indexes,which can reduce the information asymmetry existing between commercial banks and loan applicants to a certain extent so as to comprehensively evaluate the applicant's credit risk and default probability and relieve the loss caused by adverse selection and moral hazard.Second,this paper attempts to explore a more efficient personal credit rating classification model.In recent years,the ensemble learning algorithm has been proved to have better classification performance and generalization ability than the traditional single classifier,and has good performance on the classification problem.Therefore,this paper selects the latest ensemble learning algorithm–XGBoost ensemble classification algorithm as the empirical method of this paper.The Bayesian optimization algorithm is used to optimize its parameters,and the XGBoost-BOA integrated classification model is constructed.Using the credit data sets of Germany,UK and renrendai as the empirical samples,the classification performance of the XGBoost-BOA integrated classification model was tested and compared with the other eight common classification models.The empirical results show that the classification accuracy of the ensemble learning algorithm is generally higher than the single classifier classification performance;XGBoost-BOA ensemble classification model proposed in this paper has good performance in the personal credit classification of the three credit data sets,and its classification model classification accuracy is higher.In this paper,the improvement of personal credit evaluation index and the improvement of personal credit evaluation method are expected to help commercial banks quantitatively analyze the personal credit of applicants,handle loan approvals more efficiently and accurately,effectively avoid credit risk and reduce losses.It is also conducive to the improvement of the people's credit awareness and promoting the construction and development of China's credit society.This article has two innovations.First,on the basis of referring to the domestic and international personal credit evaluation standards and related scholars' research,taking its essence to its dregs,I added indicators of health status to natural conditions,added credit usage,credit amount,credit term,personal judicial record and other dishonesty records to our country's lack of credit indicators,which enriched and perfected the personal credit evaluation index system of commercial banks in our country.Second,this paper studies the application of XGBoost in the field of personal credit evaluation,and optimizes the algorithm by using Bayesian optimization algorithm to construct XGBoost-BOA.Through comparative studies,this model is validated in Germany,the United Kingdom,renrendai,which has better classification performance than other classification models.Third,this paper analyzes the importance of the feature variables of datasets in different countries by using the built-in feature importance histogram rendering function of XGBoost algorithm,which can not only effectively improve the interpretability of the model,but also help commercial banks to quickly identify the significant impact on personal credit of the characteristics of indicators.To the choice of indicators and weight settings,it also has some reference.