Enhancing Phishing Website Detection: A Comparative Analysis

In the 21st century, majority of the population in the world are connected to the Internet as the Internet allows them to do various of activities with just one click. However, with the growing prevalence of Internet of Things (IoT) devices and increased internet usage, cyberattacks are increasing in an alarming state especially phishing attacks. Hence, in this research, a hybrid approach, which combines URL-based model and content-based model, was conducted. A total of four models, Support Vector Machine (SVM), Random Forest (RF), Naïve Bayes (NB) and ensemble models were implemented in this research. Three datasets were applied, and it turned out that the proposed hybrid approach performed better than the non-hybrid approaches. The best performing model for the hybrid approach was the RF model which achieved the accuracy of 95.3% and the False Positive Rate (FPR) and False Negative Rate (FNR) of 5.4% and 3.9% respectively. Besides that, it was concluded that the ensemble model was able to detect phishing websites accurately by achieving an accuracy score of 94.7%, a FPR of 5.9% and a FNR of 4.7%.