Analysis of Machine Learning Algorithms by Developing a Phishing Email and Website Detection Model

Abstract

Machine Learning is a key branch of Artificial Intelligence that concentrates on the development of computational algorithms by creating models. It has caught major attention in the technological domain due to its various applications in speech recognition, recommendation engines, computer vision, automated stock trading etc. The model’s performance is dependent on the dataset provided and its accuracy can easily be enhanced by expanding the training dataset. Post Covid-19, it has been observed that phishing websites are appallingly on the rise, especially the phishing attacks. These attacks are caused by cybercriminals using PDF’s, Microsoft office documents and other attachments via emails. This paper focusses on discussion and comparison of different machine learning algorithms that are capable of detecting phishing emails and websites. The experiments have shown that that MultinomialNB attains the highest efficiency of 98.06% for phishing email detection and Decision Tree Classifier offers the maximum efficiency of 95.41% for phishing website detection.