Research on personal credit scoring prediction based on machine learning methods

Personal credit scoring, as a crucial component of credit risk control, has always been a hot topic in the field of financial technology. This study utilizes a dataset from a banking loan credit scoring scenario to evaluate model performance using metrics such as accuracy and precision. The objective is to develop a credit scoring model based on historical financial data. This study finds that the Lightgbm model plays an important role in feature importance ranking for personal credit scoring. Features such as "duration", "balance", and "age" significantly impact the model's performance. Lightgbm outperforms LR (Logistic Regression) and NN (Neural Network) models in terms of Precision, Recall, and F1-Score. Combining LR, NN, and Lightgbm models can achieve more stable predictive results. The equally weighted model performs well, further validating the effectiveness of model fusion. The personal credit scoring model proposed in this study can help optimize business strategies, improve credit assessment efficiency, and is particularly suitable for banks and financial institutions. It provides accurate explanations for loan application rejections and simplifies the workflow of credit professionals.