Research on the Prediction Method for Personal Loan Default Based on Two-Layer Stacking Ensemble Learning Model
Year: 2023
Type: book-chapter
Abstract: Accurate identification of loan risks to ensure the interests of financial institutions is the core of intelligent risk control.It has become an important research area to accommodate the requirements of Internet financial platforms for processing large amounts of high-latitude user data by using machine learning algorithms to build loan default prediction models.In this paper, we propose a two-layer model based on Stacking ensemble learning algorithm for personal loan default prediction, which uses LightGBM, Adaboost, XGBoost and Gradient boosting as the primary classifiers and random forest as the secondary classifier.The prediction effect of the model is verified on the personal loan default dataset of Alibaba Cloud Tianchi platform.Experimental results revealed that the Stacking ensemble learning model significantly outperforms four single algorithm model in five evaluation metrics: accuracy, precision, recall, F1 score, and AUC, with a prediction accuracy of 82.03% for the test set.Compared with the single algorithm model, the proposed Stacking ensemble learning model has better generalization ability and prediction performance in personal loan default prediction.
Authors: Zhirui Ma, Qinglie Wu
Institution: Southeast University
Language: English
Subfield: Accounting
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4385243633