An Ensemble Approach to Phishing URL Detection Using Supervised Machine Learning

Abstract

Phishing is a cybercriminal activity which deceives users into providing their private information, such as credit card credentials and sensitive information like passwords which ultimately leads to financial loss. Phishing attacks are still a serious issue in the digital world because they try to trick people into divulging important information. This study aims to address this difficulty by employing supervised machine learning to detect phishing URLs.To improve accuracy, classification methods are specifically utilized. To train and evaluate the classifiers, the study methodology makes use of a dataset that includes benign and phishing URLs along with a variety of attributes. The study compares the effectiveness of conventional single classifiers with ensemble classifiers, such as Random Forest, Gradient Boosting, and CatBoost. The experimental results show that the ensemble classifier performed better than the individual classifiers and detected phishing URLs more effectively and with a higher accuracy. The suggested method emphasizes the value of using many classifiers to increase detection accuracy and robustness, showcasing the efficiency of ensemble learning techniques in improving the classification problem. This research aims to advance cyber security measures by proving the effectiveness of ensemble classifiers in phishing URL detection and provides a dependable and effective way to counteract evolving phishing threats.