A hybrid NSGA-II and random forest framework for multi-objective credit risk assessment

Credit risk assessment is an important way to measure the likelihood of loan default. It helps institutions stay solvent and manage their capital effectively, which supports strong risk management in the financial sector. This study presents a new hybrid predictive framework that combines the Non-dominated Sorting Genetic Algorithm II (NSGA-II) with a Random Forest (RF) classifier. This approach minimizes the number of features while maximizing classification accuracy. In this setup, the decision trees in the RF model are fine-tuned by optimizing key parameters, such as the maximum number of splits and the minimum size of node partitions. Testing on the established Australian and German credit datasets shows that this framework performs at a level comparable to leading methods and traditional benchmarks, achieving classification accuracies of 93.47% and 78.00%, respectively. The resulting Pareto-optimal fronts provide risk analysts with a range of efficient solutions, allowing for clear analysis of the trade-offs between prediction accuracy and model simplicity. As a result, this research offers a strong decision-support tool that helps financial institutions reduce default risk and improve the resilience of credit portfolios. Consequently, this research provides a strong and practical decision-support framework. It helps financial institutions shift from reacting to risks to actively preventing defaults. By enabling flexible credit scoring, anticipating losses, and adjusting capital allocation based on risk, the model strengthens portfolio resilience against changes in the economy. Overall, it creates a clear, data-driven basis for modern credit governance. This reduces vulnerabilities in the system and improves risk-return balance across various lending portfolios.