Credit worthiness prediction: Approaches and methods

Abstract

This paper presents a systematic literature review on creditworthiness prediction, a critical aspect of financial risk assessment. With a focus on informing lending decisions and mitigating risks in financial institutions, the review examines methods, algorithms, and features commonly utilized in creditworthiness prediction. Through systematic searches across high-impact academic databases such as Scopus, ScienceDirect, IEEE Xplore, and SpringerLink, 25 relevant papers were collected and analyzed. Classification and regression methods emerged as predominant approaches, with Decision Trees, Random Forest, and Support Vector Machines identified as effective algorithms. Additionally, hybrid models combining traditional machine learning with deep learning techniques demonstrated promising performance. Features encompassed loan/application information, employment and income, financial history, demographics, and external factors. The findings provide insights into current practices and highlight opportunities for future research, including the integration of emerging technologies like blockchain and explainable AI, and the exploration of alternative data sources. This review contributes to advancing understanding and informs the development of more effective credit risk assessment models to support informed lending decisions and enhance financial stability.