Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{206928,
        author = {Rajan Panwar},
        title = {A Fairness-Aware Framework for Detecting and Mitigating Gender Bias in Financial Predictive Models},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {2},
        pages = {4082-4089},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=206928},
        abstract = {Algorithmic bias within automated credit scoring systems presents substantial ethical and regulatory challenges, particularly regarding gender-based discrimination, which can perpetuate systemic economic inequality. While machine learning models offer increased efficiency in risk assessment, they often inadvertently amplify historical biases embedded within training datasets. This research proposes a comprehensive, multi-layered framework for auditing and neutralizing such algorithmic bias, specifically addressing the challenges inherent in the highly imbalanced Home Credit Default Risk dataset.
We employ an eleven-feature model architecture, leveraging advanced gradient-boosting techniques to capture complex finan-cial interactions. Our methodology introduces a dual-stage miti-gation pipeline that rigorously compares pre-processing Reweigh-ing strategies with post-processing Threshold Optimization to balance fairness and utility. Our empirical analysis, validated through Stratified 5-Fold Cross-Validation, demonstrates that fairness and predictive performance are not inherently conflicting objectives; rather, they can be mutually optimized through strategic interventions.
We successfully reduced the baseline demographic disparity gap from 11.39% to an optimal 0.05%, while simultaneously enhancing the model’s overall balanced accuracy, with results confirmed as statistically significant (p < 0.05). Furthermore, we integrate SHAP (Shapley Additive Explanations) to provide robust local and global model interpretability, ensuring decision-making transparency and guarding against the emergence of hidden proxy discrimination. This research offers a scalable solu-tion for high-stakes financial environments, providing a definitive blueprint for financial institutions to align their predictive models with emerging Responsible AI mandates and rigorous regulatory compliance standards. By bridging the gap between technical performance and ethical accountability, this work contributes to the development of more equitable financial systems.},
        keywords = {Algorithmic Fairness, Demographic Parity, Bias Mitigation, Explainable AI (XAI), SHAP, Credit Risk Modeling, Proxy Discrimination, Regulatory Compliance (ECOA/GDPR), Stratified Validation, Financial Machine Learning.},
        month = {July},
        }