Credit risk assessment based on LightGBM and adaptive inertia weight PSO
In view of the complex personal credit risk problem in the loan market,the construction of credit risk as-sessment model is a very important step.Using the Lending Club dataset,the credit risk assessment model is con-structed to predict the default probability of customers.First,data processing is carried out,and then the problem of positive and negative sample imbalance is processed by the SMOTE(synthetic minority oversampling technique)al-gorithm to obtain complete credit loan data.Secondly,the LightGBM model is used for training,and the PSO(parti-cle swarm optimization)algorithm of adaptive inertia weights is used to obtain the optimal parameters of LightGBM.After comparison with multiple mainstream algorithms,experimental results show that the constructed model has bet-ter performance.
万方数据
维普
