Phishing Website Detection: An Improved Accuracy through Feature Selection and Ensemble Learning

This research focuses on evaluating whether a website is legitimate or phishing. Our research contributes to improving the accuracy of phishing website detection. Hence, a feature selection algorithm is employed and integrated with an ensemble learning methodology, which is based on majority voting, and compared with different classification models including Random forest, Logistic Regression, Prediction model etc. Our research demonstrates that current phishing detection technologies have an accuracy rate between 70% and 92.52%. The experimental results prove that the accuracy rate of our proposed model can yield up to 95%, which is higher than the current technologies for phishing website detection. Moreover, the learning models used during the experiment indicate that our proposed model has a promising accuracy rate.