Analisis Komparatif Support Vector Machine dan Random Forest untuk Deteksi Email Phishing

Abstract

Information and communication technology has rapidly advanced, bringing significant changes to daily life. With these advancements, access to information has become faster and easier; however, this convenience also introduces challenges, particularly concerning personal data security. One common cybercrime is email phishing, where attackers use malicious links to encrypt user data or devices and demand a ransom to restore access. Phishing emails often resemble official messages from trusted sources, making recipients unaware of the potential threat. To minimize such risks, technology can be utilized to automatically classify phishing emails. This study focuses on developing a machine learning model for automatic phishing email classification. The dataset used consists of 18,650 emails, including 11,322 non-phishing and 7,328 phishing emails. The proposed models employ two algorithms: Support Vector Machine (SVM) and Random Forest. To optimize performance, hyperparameter tuning was conducted using GridSearchCV. The experimental results demonstrate that the SVM algorithm achieved an accuracy of 97.27%, while the Random Forest algorithm achieved 96.51%. These findings indicate that the developed models can effectively support efforts to anticipate and mitigate phishing email threats..