Decision Tree based Framework for Business Decision Support System using Big Data Analytics

Abstract

Credit risk assessment is a critical aspect of financial decision-making, requiring accurate and efficient predictive models to evaluate a borrower’s likelihood of default. Traditional statistical methods often struggle with complex datasets, prompting the adoption of machine learning (ML) techniques for enhanced accuracy. This study explores various ML models, including Support Vector Machine (SVM), Decision Trees, Adaboost, Random Forest, K-Neighbors, Logistic Regression, and XGB Classifier, to determine the most effective approach for credit risk assessment. Extensive data pre-processing, feature engineering, and feature scaling techniques are applied to optimize model performance. The Proposed Optimized Decision Tree Classifier achieves the highest test accuracy of 83.23%, outperforming other models in predictive reliability. The research underscores the importance of selecting the right classification model for financial risk evaluation, balancing accuracy, interpretability, and computational efficiency. The results indicate that ensemble learning and hybrid approaches can further enhance prediction reliability. Future research directions include integrating deep learning techniques and real-time credit evaluation frameworks, as well as leveraging external economic indicators for a more holistic risk assessment. The findings contribute to the development of robust, data-driven strategies for financial institutions, enabling better decision-making in loan approvals and credit management.