Enhanced Phishing URL Detection through Stacked Machine Learning Model

Phishing remains a prevalent and evolving threat in cybersecurity. Phishing URL detection techniques have also continued to evolve to keep pace with the changing tactics of cybercriminals. According to the Anti-Phishing Working Group (APWG), the number of phishing attacks detected worldwide typically ranges from hundreds of thousands to millions each month. With the growing menace of phishing attacks in the cyberworld, it has become prudent to be digitally proactive. To detect phishing websites, this project proposes a robust solution for phishing website detection using a stacked classifier model that employs six algorithms: Random Forest, XGBoost, K Nearest Neighbors, Light Gradient Boosting Machine, Logistic Regression, and Support Vector Machine, keeping the Light Gradient Boosting Machine as the final estimator (or meta classifier). The stacking ensemble achieves an accuracy of 99.31%. Further, compared to stand-alone algorithms, it performs better than various parameters such as recall, precision, and f1-score at 99% each. This enables the model to be used in real-time applications as well.