Home / Current Issue / Paper 1717946
Comparative Analysis of Machine Learning Algorithms for Loan Approval Prediction
Subject area: Science,Engineering and Technology · Area of research: Machine Learning
DOI: https://doi.org/10.64388/IREV9I11-1717946
Abstract
Manual loan evaluation within the contemporary banking sector is an increasingly tedious and time-consuming process, frequently prone to human error, inconsistencies, and high operational costs. While automated systems have been introduced to improve efficiency, a significant research gap exists because many current machine learning models function as "black boxes". This lack of interpretability prevents bankers from justifying loan denials to applicants or complying with evolving "right to explanation" transparency regulations. The purpose of this study is to propose a hybrid framework that balances high predictive accuracy with fairness and explainability in credit decision-making. The methodology integrates advanced ensemble learning techniques, specifically Random Forest and XGBoost, with Explainable AI (XAI) tools such as SHAP (Lundberg & Lee) and LIME. To ensure the model remains robust against real-world data challenges, the framework utilizes SMOTE (Synthetic Minority Over-sampling Technique) for addressing dataset imbalances and the Artificial Bee Colony (ABC) algorithm for nature-inspired hyperparameter optimization. By processing diverse borrower attributes—including applicant income, credit history, and CIBIL scores—the implementation is expected to achieve a predictive accuracy exceeding 95%. Furthermore, the integration of fairness-aware learning mechanisms is projected to reduce algorithmic bias by approximately 22.4%. This provides the local and global explanations necessary for a transparent, "human-in-the-loop" financial system, shifting the primary decision-making role from a rigid algorithm back to the informed banker. Ultimately, this framework supports sustainable banking goals by minimizing non-performing assets while promoting inclusive and responsible lending practices.
Keywords
Loan Approval Prediction, Machine Learning, Classification Algorithms, Credit Risk Assessment, Comparative Analysis, Supervised Learning
How to cite this paper
@article{1717946,
author = {Maria Monish E, Dr. M N Nachappa},
title = {Comparative Analysis of Machine Learning Algorithms for Loan Approval Prediction},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {11},
pages = {2680-2687},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717946.pdf},
abstract = {Manual loan evaluation within the contemporary banking sector is an increasingly tedious and time-consuming process, frequently prone to human error, inconsistencies, and high operational costs. While automated systems have been introduced to improve efficiency, a significant research gap exists because many current machine learning models function as "black boxes". This lack of interpretability prevents bankers from justifying loan denials to applicants or complying with evolving "right to explanation" transparency regulations. The purpose of this study is to propose a hybrid framework that balances high predictive accuracy with fairness and explainability in credit decision-making. The methodology integrates advanced ensemble learning techniques, specifically Random Forest and XGBoost, with Explainable AI (XAI) tools such as SHAP (Lundberg & Lee) and LIME. To ensure the model remains robust against real-world data challenges, the framework utilizes SMOTE (Synthetic Minority Over-sampling Technique) for addressing dataset imbalances and the Artificial Bee Colony (ABC) algorithm for nature-inspired hyperparameter optimization. By processing diverse borrower attributes—including applicant income, credit history, and CIBIL scores—the implementation is expected to achieve a predictive accuracy exceeding 95%. Furthermore, the integration of fairness-aware learning mechanisms is projected to reduce algorithmic bias by approximately 22.4%. This provides the local and global explanations necessary for a transparent, "human-in-the-loop" financial system, shifting the primary decision-making role from a rigid algorithm back to the informed banker. Ultimately, this framework supports sustainable banking goals by minimizing non-performing assets while promoting inclusive and responsible lending practices.},
keywords = {Loan Approval Prediction, Machine Learning, Classification Algorithms, Credit Risk Assessment, Comparative Analysis, Supervised Learning},
month = {May},
doi = {https://doi.org/10.64388/IREV9I11-1717946}
}