<![CDATA[Internet users currently face significant risks from malicious URLs that facilitate phishing attacks, malware distribution, and data theft. Traditional blacklisting methods have become ineffective against evolving cyberattack techniques. This study proposes a Random Forest classification approach for more accurate malicious URL detection, focusing on critical URL features including URL length, presence of special keywords, subdomain structure, and special character usage. these features train the Random Forest model to distinguish between safe and malicious URLs. We evaluate model effectiveness using accuracy, precision, and recall metrics. This research aims to develop a Random Forest-based malicious URL detection system that is more accurate and adaptive than conventional methods. The study examines both the advantages and limitations of this approach, along with its potential as a reliable detection solution for dynamic digital environments. Evaluation results demonstrate an overall accuracy of 94%, weighted average F1-score of 0.94, and macro average F1-score of 0.94.]]>
Copyrights © 2025
