Credit Risk Analysis of Ecuadorian Financial Institutions: A Descriptive and Predictive Approach Integrated with Machine Learning and Business Intelligence.

Abstract

Credit risk management is essential for financial stability, especially in economies where credit access fuels growth. This study improves credit risk assessment in Ecuadorian financial institutions by integrating Machine Learning (ML), Explainable Artificial Intelligence (XAI), and Business Intelligence (BI). Using real demographic and financial data, two classification algorithms—Random Forest and XGBoost—were trained to predict loan default probability. Random Forest delivered the best results (ROC-AUC = 0.916; PR-AUC = 0.998), showing high accuracy even with imbalanced data. SHAP values explained the influence of each feature, identifying total exposure, loan maturity, and credit history as the strongest predictors. Interactive dashboards developed in Tableau converted analytical findings into practical insights for decision-makers, allowing early detection of risky clients and faster intervention. This integrated approach enhances transparency, supports proactive management, and strengthens governance in credit portfolios. The framework can be applied by other financial institutions seeking to make data-driven, fair, and efficient credit decisions.