A Comparative Analysis of Traditional Machine Learning, Deep Learning and Boosting Algorithms on Phishing URL Detection

Phishing attacks pose a significant threat to user’s data security by using counterfeit Uniform Resource Locators (URLs) that are hard to distinguish from legitimate ones. Machine learning can help identify phishing URLs by analyzing their features. This research work compares phishing detection mechanisms using Machine Learning, Deep Learning, and Boosting techniques, evaluating their accuracy across five datasets. The datasets are: Dataset-1 (Kaggle, 2 class), Dataset-2 (UCI, 2 class), Dataset-3 (Mendeley, 2 class), Dataset-4 (UNB, 5 class) and Dataset-5 (Kaggle, 4 class). The analysis shows that boosting techniques outperform traditional Machine Learning and Deep Learning, achieving detection accuracies of $\mathbf{9 7 \%, 9 6 \%, 9 8 \%, 9 8 \%}$ and 92% across the five datasets respectively.