An Explainable AI Model for Credit Evaluation and Loan Decision Support
ID：117
View protection：Participant Only
Updated time：2026-07-22 16:10:08 Views：14
Online
Abstract
Loan approval is a core function of banking because it impacts institutional profitability, credit accessibility, and financial stability. While artificial intelligence and machine-learning models can improve the speed and accuracy of credit evaluation, many high-performing models are difficult for customers, credit officers, and regulators to interpret, Such lack of transparency is problematic if an application is rejected as applicants need comprehensible reasons and lenders need to demonstrate consistency, fairness and policy compliance,. In this paper, an explainable artificial intelligence (XAI) framework for credit evaluation based on the Logistic Regression, Support Vector Machine, Decision Tree and Random Forest classifiers is proposed. We evaluate the framework on a public loan-approval dataset and explain it using LIME, SHAP and Partial Dependence Plots. Random Forest model gave the best reported performance with accuracy, sensitivity and specificity of 0.998, 0.998 and 0.997 respectively. The explanatory layer is useful in explaining local decisions for specific applicants as well as global features effects over the data set. The proposed framework offers a transparent, accountable, and customer-centric approach to credit decision-making. Prior to practical implementation, additional validation, fairness testing, leakage checks, and regulatory assessment are required.
Keywords
Index Terms— Explainable AI; Credit Risk; Loan Approval; Random Forest; LIME; SHAP; Partial Dependence Plot; Machine Learning in Finance.
Post comments