Machine Learning-Based Prediction of Credit Default Risk and Profiling of High-Risk Users in Buy Now, Pay Later (BNPL) Services

Xinru Lu

doi:10.54097/10vr0a69

Authors

Xinru Lu School of Shanghai University of Finance and Economics, Shanghai, China

DOI:

https://doi.org/10.54097/10vr0a69

Keywords:

BNPL, credit risk, default prediction, machine learning, user profiling, association rules.

Abstract

This study looks at BNPL credit risk using Kaggle data. We use several machine learning models, along with clustering and association rules, to predict defaults, group users, and find risk patterns. Logistic regression works best among the models. Monthly income predicts default better than debt ratios; this shows the cash‑flow limits caused by BNPL's zero down payment, small orders, and weak credit records. Clustering splits users into three groups, with low‑income and high‑debt users showing the highest risk. Association rules find high‑default trait pairs like “low income + young” and “high debt + young”. These results offer useful guidance for BNPL platforms in risk checks, user grouping, and different risk controls.

References

[1] Di Maggio M, Williams E, Katz J. Buy Now, Pay Later Credit: User Characteristics and Effects on Spending Patterns[R]. NBER Working Paper No. 30508, 2022. DOI: 10.3386/w30508.

[2] Hayashi F, Routh A. Financial Constraints Among Buy Now, Pay Later Users[J]. Economic Review, 2025, 110(4): 1-18. DOI: 10.18651/ER/v110n4HayashiRouth.

[3] Liang H C. The Construction of Legal Mechanism for Consumer Over-Indebtedness Risk and Rights Protection under “Buy Now, Pay Later” Model[J]. E-Commerce Letters, 2025, 14(10): 397-404. DOI: 10.12677/ecl.2025.14103158.

[4] Jose A, Kelly J. Who Clicks “Pay Later”? Financial Vulnerability and Buy Now Pay Later Usage[R]. Central Bank of Ireland Staff Insights, No. 7/SI/25, 2025.

[5] Zhang J, Liu Y. Credit-card consumer-credit default-risk identification and credit-scoring model[J]. Theory and Practice of Finance, 2023(7): 89-97.

[6] Ma Y C, Hong H Q, Lin L Q. Reverse incentives for borrower performance in online lending: Empirical evidence from Renrendai platform[J]. Journal of Finance and Economics, 2020, 46(5): 66-80.

[7] Zhang Y W, Gao W H. Self-construals, Cultural Difference and Credit Risk: Evidence from Internet Finance[J]. Journal of Finance and Economics, 2020, 46(1): 34-48. DOI: 10.16538/j.cnki.jfe.2020.01.003.

[8] Li Y, Liu L. Reconstructing consumer-credit evaluation framework in digital era[J]. Journal of Financial Research, 2021(11): 1-18.

[9] Li J J, Wang S Z. Research on user credit risk-evaluation system under consumer-finance scenarios[J]. Theory and Practice of Finance, 2022(5): 45-51.

[10] Hu J, Li Q, Zeng Y, Tian X, Liu S. Digital footprints and default prediction: Evidence from consumer credit[J]. Journal of Management Sciences in China, 2025, 28(2): 115-139.

[11] Ayari H, Guetari R, Kraïem N. Machine learning powered financial credit scoring: A systematic literature review[J]. Artificial Intelligence Review, 2026, 59: Article 13. DOI: 10.1007/s10462-025-11416-2.

[12] Sun L, Wang X Y, Ma R X. Bad-user profiling for micro-loan based on Logistic model[J]. Statistics and Applications, 2023, 12(6): 1696-1706.

[13] Dai W S, Liu Z X. Research on big-data credit assessment in consumer finance: Based on GBDT and XTBoost[R]. IMI Working Papers No. 1925, International Monetary Institute, Renmin University of China, 2019.

[14] Ghoujdam M E K, Chaabita R, Elkhalfi O, et al. Consumer credit risk analysis through artificial intelligence: A comparative study between the classical approach of logistic regression and advanced machine learning techniques[J]. Cogent Economics & Finance, 2024, 12(1): 2414926. DOI: 10.1080/23322039.2024.2414926.

[15] Khalili P, Kargari M, Rastegar M A, et al. Predicting BNPL Loan Defaults: A Comparison of Ensemble Learning Models Combined with Balancing Techniques and an Analysis of the Impact of Digital Literacy[J]. Engineering Reports, 2026, 8(2): e70601. DOI: 10.1002/eng2.70601.

[16] Taş C. Comparison of Machine Learning and Standard Credit Risk Models’ Performances in Credit Risk Scoring of Buy Now Pay Later Customers[D]. Ankara: Middle East Technical University, 2023.

[17] Mahmud M R, Hoque M R, Ahammad T, et al. Advanced AI-Driven Credit Risk Assessment for Buy Now, Pay Later (BNPL) and E-Commerce Financing: Leveraging Machine Learning, Alternative Data, and Predictive Analytics for Enhanced Financial Scoring[J]. Journal of Business and Management Studies, 2024, 6(2): 180-189. DOI: 10.32996/jbms.2024.6.2.19.

[18] Mishra A, Mou S N, Ara J, et al. Regulatory and Ethical Challenges in AI-Driven and Machine Learning Credit Risk Assessment for Buy Now, Pay Later (BNPL) in U.S. E-Commerce: Compliance, Fair Lending, and Algorithmic Bias[J]. Journal of Business and Management Studies, 2025, 7(2): 42-51. DOI: 10.32996/jbms.2025.7.2.3.

[19] Yuan Y, Yang Y, Wang H Q. Research on Credit Risk Assessment of P2P Lending Platform[J]. Management Science and Engineering, 2020, 9(3): 157-164. DOI: 10.12677/mse.2020.93021.

[20] Tanantong T, Loetwiphut P. Association rule mining framework for financial credit-risk analysis in peer-to-peer lending platforms[J]. Science, Engineering and Health Studies, 2023, 17: 23020006. DOI: 10.69598/sehs.17.23020006.