Optimizing Phishing Detection Systems with Ensemble Learning: Insights from a Multi-Model Voting Classifier

Phishing attempts threaten cybersecurity, hence strong detection systems are necessary. This research study highlights the utilization of ensemble learning methods to improve phishing detection classifications. This research focuses on increasing the detection accuracy, precision, recall, and F1-score by integrating Random Forest, Support Vector Machine, and CatBoost models into a Voting Classifier. The Voting Classifier outperformed all the individual models’ accuracies by 97.84%. Class-0 also showed better accuracy (0.9784) and recall (0.9784) coupled with a balanced F1-score of 0.9725. By properly using the benefits of every base model, the combined strategy reduces its shortcomings and offers a more robust detecting system. These results show how well ensemble learning could support phishing defences and forward cybersecurity. This work highlights the need to combine many machine learning methods to create strong, highly-performing phishing detection systems.