Classification of Phishing Website Using Machine Learning Based Proposed Ensemble Model

Abstract

Phishing is the process of trying to get sensitive data from unauthorized persons, such as usernames, passwords, credit card numbers, and debit card information. Since there is no one method to properly reduce every vulnerability due to the problem of phishing. There are many techniques are frequently utilized to reduce specific attacks. The main aim of this research work is to develop a robust machine learning based model that can detect and classify URL based phishing data. This paper proposed a new ensemble model that is a combination of Decision Tree (DT), Support Vector Machine (SVM), and Logistic Regression (LR) using the voting scheme ensemble technique. The proposed ensemble model compared with other individual classifiers like Decision Tree (DT), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Gaussian Naive Bayes (GNB), Logistic Regression (LR) techniques, and ensemble classifiers like Extreme Gradient Boosting (XGBoost), Adaptive Boosting (AdaBoost) where our proposed ensemble model achieved remarkable performance of 99.02% accuracy with 10-fold cross validation technique. Finally, machine learning based classification techniques are suitable and robust methods that handle the dynamic nature of phishing techniques and provide an accurate method of classification.