Comprehensive Literature Survey on Machine Learning for Credit Scoring of Thin File Consumers

The proliferation of machine learning (ML) has ushered in significant advancements in credit scoring, particularly for thin-file consumers-individuals with limited credit histories who are often underserved by traditional scoring models. This literature survey comprehensively reviews recent developments in ML algorithms and their application in credit scoring. It delves into the superior performance of neural networks, support vector machines, and ensemble methods like random forests and gradient boosting machines. Additionally, the survey addresses the critical issues of bias and fairness, leveraging alternative data sources, and the importance of interpretability and explainability in ML models. By synthesizing current research, identifying gaps, and proposing future directions, this survey aims to contribute to the development of more accurate, fair, and interpretable credit scoring models, thereby enhancing financial inclusion for thin-file consumers.