Predictive risk assessment credit scoring using supervised learning

Khor, Wei Heng (2025) Predictive risk assessment credit scoring using supervised learning. Final Year Project, UTAR.

Preview

PDF Download (6Mb) | Preview

Abstract

This study explores the application of supervised learning models within credit scoring, aiming to revolutionize risk assessment in lending decisions. The primary goal involves comparing these advanced methodologies against conventional credit assessment techniques to ascertain their effectiveness in determining creditworthiness. In response to the escalating complexity of financial transactions and the wealth of available data, this research seeks to elevate the precision and efficiency of credit risk evaluation. Supervised learning, known for its ability to learn from labelled datasets, presents an opportunity to redefine credit scoring by leveraging historical credit information. The core focus is on assessing the predictive capabilities of supervised learning algorithms—specifically Logistic Regression, Random Forest, K-Nearest Neighbours, Support Vector Machines and Gradient Boosting—against established credit scoring methods. By harnessing the power of these modern techniques and analysing intricate credit patterns, this research endeavours to deliver more accurate credit risk assessments. It strives to surpass the existing industry norms by using machine learning models to refine credit evaluation processes. Beyond academia, this study aims to introduce substantial advancements in credit risk assessment methodologies. It seeks to bridge the gap between conventional and contemporary approaches by revolutionizing credit scoring. By tapping into supervised learning's potential, this research aspires to produce predictive credit scoring models that surpass industry standards, fostering more reliable lending decisions. The objectives encompass the development of more accurate credit scoring models, identification of influential creditworthiness factors, and the enhancement of model interpretability. The fusion of traditional credit assessment wisdom with the cutting-edge capabilities of supervised learning intends to empower financial institutions with sophisticated tools for making informed, expedited, and more reliable lending decisions. Ultimately, this research aspires to create a paradigm shift in credit risk assessment, enabling financial entities to navigate evolving market conditions with confidence and precision.

Actions (login required)