Phishing Website Detection Using Machine Learning

Abstract – Phishing is a deceptive technique used by cybercriminals to impersonate legitimate websites and obtain sensitive user information such as login credentials and financial data. This project proposes a machine learning-based approach to detect phishing websites by analyzing URL features and content-related indicators. Various algorithms, including Logistic Regression, Multinomial Naïve Bayes, and XGBoost, were evaluated for their classification accuracy. The system employs preprocessing techniques such as tokenization, stemming, and vectorization using Count Vectorizer to convert textual URL data into machine-readable formats. FastAPI is utilized for seamless deployment of the trained model, providing a user-friendly interface. Among the tested models, Logistic Regression achieved the highest accuracy of 96.63%, demonstrating its effectiveness in phishing detection. The proposed solution not only enhances detection accuracy but also enables real-time identification of phishing threats, contributing to a safer online environment. Key Words: Phishing Detection, Logistic Regression, XGBoost, Multinomial Naïve Bayes, URL Features, FastAPI, Machine Learning.