Machine Learning Approaches for Phishing Website Detection: A Comparative Analysis

Phishing attacks are a major threat to cyber security because they exploit the trust millions of users have in seemingly legitimate websites to gain access to their personal information. Machine learning algorithms offer methods for the detection of phishing websites by analyzing the features and patterns that distinguish them from legitimate phishing websites. The aim of this study was to address the issue of phishing recognition by providing a precise verification of websites through modern machine learning techniques. In this research work phishing data were collected using Google's search operators, the PhishTank archive, and the UCI Machine Repository MillerSmiles archive. Collected dataset were processed for the missing values, followed by model generation using R studio, different machine learning models were created using supervised machine learning models; Decision Tree (DT), Support Vector Machine (SVM), Logistic Regression (LR), Naïve Bayes (NB), and Random Forest (RF). Results shows that the RF model had the best Results with accuracy of 97%. The DT was closely behind with 95%, with LR and SVM at 94.4% and 93% respectively. The NB model was also only yielding an accuracy of 87.2% which made it not well suited for this application. The algorithm has compared on the basis of various parameters like accuracy, precision, recall, F-measures, error rate, etc. and in all terms the RF performed better than others.