Personal Credit Loans Risk Prediction Based on NS3-LightGBM

In recent years, the demand for personal credit loans has been increasing day by day. For banks, how to accurately identify and effectively predict whether borrowers will repay on time is a highly concerning issue. To effectively address a series of key problems existing in traditional loan risk prediction models, such as insufficient prediction performance, single hyperparameter optimization objectives, and poor model interpretability, this research integrates machine learning algorithms such as LightGBM and NSGA-III and builds an algorithm called NS3-LightGBM for predicting the probability of borrowers repaying on time. Testing and empirical analysis are done to confirm the suggested model’s ability to make predictions. The suggested model has an accuracy rate of more than 86%, according to the results. in prediction, and its prediction ability is better than traditional machine learning models. In addition, through a more detailed analysis of the impact of various features on the prediction results, it is found that monthly income, the number of months on-time repayment per year, and whether there is a fixed job are key features that affect the possibility of borrowers repaying on time.