Research On Personal Credit Evaluation Model Based On LightGBM Algorithm

In recent years,China has accelerated the construction of social credit system,and the awareness of social credit has been constantly enhanced.The demand of citizens for understanding personal credit situation is also increasing.Personal credit evaluation problem has gradually become a hot research topic in today's digital world.However,the research on personal credit assessment started late in China,which has a certain gap with developed countries.And a large number of domestic financial institutions lack of algorithm model which can accurately evaluate personal credit,so the research on personal credit evaluation model is urgent.LightGBM algorithm is an efficient integrated learning algorithm with the advantages of extreme gradient lifting and parallel computing.In this paper,based on LightGBM algorithm,a personal credit evaluation model is established.and using the idea of integrated learning to study the evaluation and prediction of personal credit.According to the real transaction data of lending Club platform,we carry out empirical research from data preprocessing,feature engineering,model training and prediction,and finally make comparative analysis with GBDT and XGBoost models.The results show that LightGBM algorithm can not only ensure the classification effect,but also improve the operation efficiency to a certain extent.Therefore,LightGBM algorithm is a very valuable integrated learning algorithm in personal credit evaluation.