Phishing Website Detection using Dimensionality Reduction and Machine Learning

Phishing attacks have become an increasingly prevalent threat, with scammers masquerading as trustworthy entities to deceive victims into divulging sensitive information such as usernames, passwords, and credit card details. These malicious attempts have resulted in significant financial losses and compromised personal data for countless individuals. In response to this growing concern, our research aims to develop a robust anti-phishing website framework. To achieve this, we employed a range of supervised machine learning algorithms to distinguish between legitimate and phishing websites. Our approach combined the strengths of Support Vector Machine (SVM), Decision Tree, Random Forest, and XGBoost classifiers to create a hybrid detection model. By analyzing the characteristics of phishing and legitimate websites, we identified key features that enabled our model to accurately detect phishing attempts. Our evaluation of various feature selection-classification algorithms revealed that our hybrid technique achieved an impressive accuracy rate of 98.28%. This result demonstrates the effectiveness of our approach in detecting phishing websites, positioning it alongside state-ofthe- art techniques in this field.