Efficient Malicious Website Detection Technique Using Stacking Ensemble of Machine Learning Models

This paper analyzes the capability of various machine learning (ML) algorithms to identify and thereby avoid phishing attacks. As a form of cyberattack, phishing directs users to fake websites and tricks them into revealing sensitive personal details, such as account passwords, bank account information, and ATM pin-card details. Our proposed malicious website classification system extracts features to counter common phishing detection approaches and utilizes classical ML algorithms, such as logistic regression (LR), K-nearest neighbors (K-NN), random forest (RF), support vector machine (SVM), Naive Bayes classifier, and decision tree for classifying the legitimacy of a website. We have also developed a stacked model of three selected models to determine the maliciousness of a website. We check the efficiency of various ML techniques in detecting malicious websites. The proposed stacked model proved to be efficient by achieving 0.969 accuracy, 0.972 precision, 0.975 recall, and 0.969 F1 score.