Sanket Lodha, Nishikant Patil, Aditya Gaikwad, Ashutosh Misar

DOI: https://doi.org/10.5281/zenodo.24198902

Credit risk prediction is a critical component of banking and the financial sector. It allows predicting whether the borrower is likely to repay the credit. Conventional credit risk assessments rely on credit scores, credit history, income, and already-existing credits. But a considerable number of people including students, entrepreneurs, and those who have a limited credit history might lack the financial data required for proving their creditworthiness. Alternative data collection can resolve this problem to some extent by collecting more data on the borrower’s financial behaviour. Such type of data can include payment of bills or rent, payment transaction history, online payments, and any other financial activities of a client. The machine learning approach helps analyse all these data types in order to find patterns useful in improving the prediction. This research paper analyses the application of alternative data for the prediction of credit risk, main studies in this area, possible pros and cons of using this approach. It will also provide an overview of the most critical issues related to the use of alternative data, such as privacy issues, algorithmic discrimination, quality problems, and lack of interpretability.

Credit Risk Prediction, Alternative Data, Machine Learning, Financial Inclusion, Credit Scoring, Data Privacy.

Predicting credit risk is one area of the financial field where data analytics can help make accurate predictions. The use of traditional techniques of credit scoring can prove to be efficient, but they might fail to give a holistic view of those people who do not have a sufficient credit record. Information such as utility bill payments, rental bill payments, and other digital financial transactions could give valuable insight into evaluating those customers. The literature covered in this paper gives clear indications about the possibilities offered by machine learning and alternative data. But at the same time, success depends upon several other aspects as well. Privacy, algorithmic bias, explain-ability, cyber-security, and regulation compliance continue to be issues of significance. A model cannot be regarded as a successful one only based on its superior performance. Rather, it should ensure reliable results and maintain fairness as well as privacy of the borrower. Alternative data can facilitate credit risk prediction through a combination of alternative and traditional data as well as proper measures. Further research may concentrate on the development and testing of such a model by utilizing a publicly available credit risk dataset. The comparison of traditional and machine learning models as well as their predictive performance and fairness will be necessary.

1.Hand, D. J., & Henley, W. E. (1997). Statistical classification methods in consumer credit scoring: A review. Journal of the Royal Statistical Society: Series A (Statistics in Society), 160(3), 523–541.
2.Lessmann, S., Baesens, B., Seow, H.-V., & Thomas, L. C. (2015). Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research, 247(1), 124–136.
3.Jagtiani, J., & Lemieux, C. (2019). The roles of alternative data and machine learning in fintech lending: Evidence from the LendingClub consumer platform. Financial Management, 48(4), 1009–1029.
4.Berg, T., Burg, V., Gombović, A., & Puri, M. (2020). On the rise of FinTechs: Credit scoring using digital footprints. The Review of Financial Studies, 33(7), 2845–2897.
5.Fuster, A., Goldsmith-Pinkham, P., Ramadorai, T., & Walther, A. (2022). Predictably unequal? The effects of machine learning on credit markets. The Journal of Finance, 77(1), 5–47.

Sanket Lodha, Nishikant Patil, Aditya Gaikwad, Ashutosh Misar (2026). Credit Risk Prediction Using Alternative Data. International Journal of Computer Techniques, 13(6), 766–773. ISSN: 2394-2231. DOI: https://doi.org/10.5281/zenodo.24198902