Romanian Cyber Security Journal / Fall 2025, No. 2, Vol. 7

A Study on Detecting and Preventing Phishing Attacks Using Machine Learning Techniques

Ana-Maria BROȘTIC, Bianca-Andreea MĂRGĂRIT, Diana-Sînziana MIHAI, Andreea NICOLESCU, Emil SIMION

---

Abstract

Phishing is a major cybersecurity threat that targets users and organizations by exploiting deceptive e-mail tactics to steal sensitive data. Because the traditional methods often fail against the evolving phishing attacks, the authors explore how machine learning approaches can improve phishing detection. They evaluate the supervised learning models, including Support Vector Machine, Logistic Regression, and Random Forest, comparing their accuracy, precision, and sensitivity. The study also employs feature selection, data preprocessing, and ensemble learning to enhance the results. The goal is to improve user protection by identifying the most effective machine learning-based solution for detecting phishing attacks and advancing cybersecurity defense strategies.

Keywords

Machine learning, Cybersecurity, Phishing, Support Vector Machine, Random Forest, E-mail, Logistic Regression, User protection, Phishing attacks