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{207758,
        author = {Nirmala M S and Siddaraju K and Madhura Yadav  M.P and Anil Kumar R J},
        title = {An Enhanced Hybrid Feature Selection and Ensemble Learning Framework for Early Detection of Phishing Websites},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {3},
        pages = {2758-2770},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=207758},
        abstract = {Phishing attacks remain one of the most prevalent cyber threats, causing significant financial and reputational damage to individuals and organizations worldwide. Traditional detection methods often struggle with evolving attack patterns and high-dimensional feature spaces. This research proposes an enhanced hybrid framework combining advanced feature selection techniques with ensemble learning algorithms for early detection of phishing websites. We employ a three-stage hybrid feature selection strategy integrating filter-based methods (Mutual Information), wrapper-based techniques (Recursive Feature Elimination with Random Forest), and embedded approaches (Feature Importance). The selected features are processed through a sophisticated ensemble learning architecture comprising XGBoost, LightGBM, Random Forest, Gradient Boosting, and Support Vector Machines combined via stacking and voting mechanisms. We evaluate our framework on the UCI Phishing Websites dataset containing 11,055 instances with 30 features. Experimental results demonstrate that our hybrid approach achieves 97.8% accuracy, 97.5% precision, 98.1% recall, and 97.8% F1-score, significantly outperforming traditional single-classifier methods. The three-stage feature selection reduces dimensionality by 33% while maintaining classification performance. Cross-validation analysis confirms the model's robustness and generalization capability. Our findings indicate that integrating multiple feature selection paradigms with ensemble learning creates a more resilient framework for phishing detection, offering substantial improvements for cybersecurity applications and real-time threat mitigation systems.},
        keywords = {Phishing Detection, Ensemble Learning, Hybrid Feature Selection, Cyber Security, Machine Learning, XGBoost, Random Forest, Recursive Feature Elimination},
        month = {August},
        }