Abstract

Phishing emails pose a severe risk to online users, necessitating effective identification methods to safeguard digital communication. Detection techniques are continuously researched to address the evolution of phishing strategies. Machine learning (ML) is a powerful tool for automated phishing email detection, but existing techniques like support vector machines and Naive Bayes have proven slow or ineffective in handling spam filtering. This study attempts to provide a phishing email detector and reliable classifier using a hybrid machine classifier with term frequency-inverse document frequency (TF-IDF) and an effective feature extraction technique (FET) on a real-world dataset from Kaggle. Exploratory data analysis is conducted to enhance understanding of the dataset and identify any conspicuous errors and outliers to facilitate the detection process. The FET converts the data text into a numerical representation that can be used for ML algorithms. The model’s performance is evaluated using accuracy, precision, recall, F1 score, receiver operating characteristic (ROC) curve and area under the ROC curve metrics. The research findings indicate that the hybrid model utilising TF-IDF achieved superior performance, with an accuracy of 87.5%. The paper offers valuable knowledge on using ML to identify phishing emails and highlights the importance of combining various models.