Unlocking Loan Approval: Data-Driven Insights and Machine Learning Precision

Abstract

This research study explores the realm of loan approval prediction, leveraging a comprehensive dataset encompassing a wide array of applicant-related variables, including loan parameters, credit scores, education, and asset holdings. The primary objective is to construct an effective predictive model to aid lending institutions in making informed decisions regarding loan applications. Furthermore, this study identifies key factors that significantly influence loan approval determinations, offering insights into prioritising services for applicants with higher approval probabilities. The dataset employed in this study comprises critical financial information typically used to assess loan eligibility, such as CIBIL scores, income, employment status, loan terms, loan amounts, asset values, and loan status. Employing advanced machine learning and data analysis techniques, this research develops predictive models capable of estimating the likelihood of loan approval based on these features. Key findings reveal that a higher CIBIL score positively correlates with increased chances of loan approval, while a larger number of dependents diminishes approval probabilities. Additionally, applicants with more substantial assets, encompassing both movable and immovable holdings, are more likely to secure loan approval. Furthermore, individuals requesting higher loan amounts with shorter tenures exhibit greater odds of approval. The research employs machine learning models, including the Decision Tree Classifier and Random Forest Classifier, to forecast loan approval outcomes. These models yield impressive accuracies of 92.98% and 90.98%, respectively, with the Decision Tree Classifier outperforming the Random Forest Classifier within this context.

DOI: https://doi.org/10.21863/ijbri/2024.12.2.008

Keywords: Loan Approval Prediction, Credit Risk Assessment, Machine Learning Models, Financial Decision-Making, Lending Institutions

View PDF