Phishing Email Classification Using Machine Learning

This study presents a comparative analysis of classical machine learning classifiers for phishing email detection. Using a public dataset of 18,570 emails, we evaluate six widely used algorithms: Multinomial Naive Bayes, Logistic Regression, Support Vector Classifier (SVC), Decision Tree, Random Forest, and Gradient Boosting. The email text is preprocessed using standard NLP techniques (lowercasing, punctuation removal, tokenization, stop-word removal, and Porter stemming) and represented using Term Frequency-Inverse Document Frequency (TF-IDF) weighting. Models are evaluated using accuracy, precision, recall, F1-score, and confusion matrices. The study also includes an exploratory analysis of the corpus, highlighting class-indicative terms, and discusses key limitations, such as the inadvertent removal of negation words by standard stop-word lists and the presence of HTML artifacts. The complete source code and dataset are openly available to ensure reproducibility.