Enhanced Phishing Detection: Integrating Random Forest Classifier and Domain Analysis for Proactive Cybersecurity
N. Dharini, E P Vishnu Sudarsan, Bhavesh Praveen, D Thirugnanam, K Sibi · 2024
This work provides a multimodal strategy for detecting phishing attack that combines URL structure analysis and machine learning methodologies. The process starts with Level 1 analysis, which includes parsing the URL components, accessing a list of known phishing sites, and evaluating domain-specific features to detect irregularities. Following Level 1 analysis, a decision is made whether to proceed to Level 2. Level 2 involves loading the URL into a machine learning model to detect specific patterns and features that signal phishing attempts. Our suggested technique combines Level 1 and Level 2 assessments to provide a complete assessment of the URL's legitimacy and possible phishing activities. Using the Random Forest Classifier, we achieved a maximum accuracy of 96% based on experimental verification which outperforms other popular classifiers including XGBoost, Decision Tree Classifiers, Logistic Regression, Bagging Classifier, AdaBoost Classifier, and Gradient Boosting Classifier. Performance parameters such as accuracy, recall, and F1-score were calculated and compared.