Phishing URLs Using Machine Learning Hybrid Stacking Classifier Approach with XGBoost, Random Forest and Extra Trees

Within the dynamic field of cybersecurity threats, phishing remains a prevalent and sophisticated tactic employed by malicious actors to deceive consumers. This paper presents a novel approach that leverages the strengths of Random Forest, ExtraTrees, and XGBoost in a machine learning framework to enhance the detection of phishing URLs through a hybrid stacking classifier. The suggested method combines the best features of all three ensemble learning techniques to provide a phishing detection model that is reliable and precise. An ensemble of classifiers is formed by combining XGBoost, which is well-known for its boosting abilities, with the highly randomized ExtraTrees method and the adaptable Random Forest algorithm. By using the various decision limits of each base learner, the hybrid stacking model combines their particular capabilities in the best possible way to provide higher prediction performance. Preprocessing is done on the training dataset to extract pertinent characteristics such as content, URL structure, and contextual data. The efficacy of the suggested method in detecting phishing URLs is assessed by extensive testing on an extensive dataset. The results indicate how well the hybrid stacking classifier performs in comparison to separate classifiers and highlight the possibility for improved phishing detection in practical situations. The study has importance as it presents a novel and efficient machine learning-based approach to counteract the rising danger of phishing assaults, hence propelling the area of cybersecurity forward.