Abstract

This research presents a complete cybersecurity solution, not only for phishing URL detection but also for malware analysis, through two methods: signature-based matching and machine learning techniques. For phishing URL detection, a signature-based method compares URLs against several known phishing datasets, while a Random Forest-based machine learning model analyzes 30 specific features of URLs to attain a high accuracy percentage. The malware analysis module employs the dual modality of generating SHA-256 hashes to match against signatures in malware databases, while Random Forest classification is also used for detecting activities representing signs of malicious behavior. Random Forest was selected for use in both due to its proven track record in cybersecurity tasks, achieving up to 99.36% accuracy in detecting phishing incidents and about 99% accuracy when classifying malware. The system fills existing gaps in current cyber defenses, combining traditional signature-based methods with more advanced machine learning to bring realistic protection against both known and emerging threats. This hybrid method has enormous advantages over pure single-method approaches, especially in detecting zero-day attacks and polymorphic malware that evade signature detection alone.