Implementation of the Random Forest Algorithm for Loan Eligibility Prediction and Feature Analysis Based on Financial Data

Abstract

The advancement of information technology has led to an increasing demand for loan access, both through banking institutions and online lending platforms. However, the process of evaluating loan eligibility, which is still carried out manually or semi-manually, is prone to human error and decision-making bias, ultimately increasing the risk of loan defaults. This study aims to implement the Random Forest algorithm to predict loan eligibility based on financial data, as well as to evaluate its accuracy. The dataset used in this study is loan_approval_dataset.csv, which is downloaded from Kaggle, utilizing 11 input features. The system is developed as a web-based application using Laravel as the main frontend and backend framework, while Flask is used as a backend API for executing the machine learning processes. The testing results show that the Random Forest model achieves an accuracy of 98.44%, with a precision of 98.14%, recall of 99.37%, and an F1-score of 98.75%. Furthermore, the cibil score feature is identified as the most influential factor in the prediction process, contributing 80.65% to the model's outcome. These findings indicate that the Random Forest algorithm is highly effective for use in a loan eligibility prediction system, as it provides fast, objective, and highly accurate results.