Synergistic Machine Learning and Bio-Inspired Optimization for Precision Credit Risk Modeling: A Systematic Review
Authors: AYODELE Emmanuel Oladimeji, DAKOH Wisdom, EKPE Marvellous Solomon, AKINTUNDE Adedamola Emmanuel, IDOWU Peter AdebayoThis systematic review provides a comprehensive analysis of the evolving landscape of consumer credit risk modeling, focusing on the synergistic integration of decision tree algorithms and bio-inspired feature selection metaheuristics. As financial institutions transition toward automated underwriting, the need for transparent, high-precision classification models becomes both an operational and regulatory necessity. This paper evaluates the strengths and limitations of traditional statistical models, ensemble methods, and deep learning architectures, detailing the theoretical underpinnings of inherently interpretable decision trees. Furthermore, it examines data engineering strategies; comparing label and one-hot encoding and investigates mechanisms for dimensionality reduction. Specifically, it contrasts information-theoretic filter methods like Mutual Information with wrapper-based swarm intelligence techniques, such as Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO). By analyzing the extraction of top-k feature subsets, the study highlights the methodology for identifying a "Stable Core" of features essential for robust decision support systems. Finally, this review synthesizes critical contemporary trends, including Agentic AI, Explainable AI, and alternative data integration, while outlining current research gaps and charting a clear path toward autonomous, fair, and responsible financial AI ecosystems.

