Abstract

Cybercriminals use special ways of writing, unique structures, and urgent language to trick people into falling for their scams. Traditional methods for catching phishing emails, like checking for certain keywords or filtering content, aren’t working as well anymore because hackers are always changing how they attack. In this project, a machine learning model was built to help identify phishing emails on the web. This model looks at how emails are written, their purpose, and how they interact with users, which is a way of analyzing behavior. The Kaggle Phishing Emails Dataset was used to build and compare models that can tell the difference between normal emails and those sent by scammers. Random Forest, Support Vector Machine, and Logistic Regression models were tested to find the best one. The results showed that Logistic Regression was the most accurate and reliable for this kind of analysis, offering a practical foundation for adaptive, behavior-based phishing threat mitigation.