Revolutionizing Loan Approval: Harnessing the Power of K-Nearest Neighbors for Predictive Eligibility

A bank has become a basic need of our daily lives. People use banks for several tasks, such as depositing or withdrawing money and borrowing loans. The number of loan applications is rising significantly. It is crucial for the bank and financial sectors to determine loan eligibility and approve loans to suitable applicants only. This study attempts to design a system which can easily predict loan eligibility with the application of several machine learning (ML) algorithms. We employed the five different machine learning algorithms to achieve this research’s aim. These algorithms were trained using the pre-processed dataset to create predictive models. F1 score, accuracy, Recall and Precision are used to evaluate the performance of the proposed approaches. The result showed that K-nearest neighbor (KNN) demonstrated outstanding performance, with an accuracy rate of 88.89%. The Random Forest (RF) model had an accuracy of 84.44%, whereas the Support Vector Machine (SVM) and Linear regression (LR) models achieved 82.22% and 80% accuracy, respectively. The Decision Tree (DT) algorithms showed the worst performance among the five algorithms, with an accuracy of 79.88%. The achievement of this research focuses on the possibility of machine learning algorithms as an efficient method for predicting loan eligibility in banks, contributing to the advancement of financial technology by automating and optimizing key banking processes, reducing processing times, and enhancing overall bank efficiency.