Phishing Detection: Performance Evaluation of Both Ensemble and Classical Machine Learning Models

Abstract

The majority of bank customers have switched to e-banking for their regular financial activities as a result of the development of technology and the rise  of electronic commerce in global trade. The  emergence of e-commerce has also attracted  scammers looking to swindle bank customers,  rendering e-commerce platforms vulnerable to a  number of assaults, the most frequent of which is  phishing. The application of ensemble learning and  classical machine learning techniques for the  detection of phishing on e-commerce websites is   therefore explored in this study, and the  performance of the models is further evaluated. In  order to learn phishing patterns from an   e-commerce website phishing dataset, three  ensemble learning algorithms—adaptive boosting,  majority voting, and stacking  ensemble—were used.  Also included are three classical machine learning  classifiers: Naive Bayes (NB), K-Nearest Neighbour  (KNN), and  Decision Tree (DT). The created models  were utilized to find cases of phishing in new,  previously undiscovered data. According to the   study's findings, Naive Bayes, K-Nearest Neighbour,  and Decision Tree all had accuracy rates of 89.69%,  93.34%, and 83.90%, respectively.  Moreover, the  voting ensemble classifier's accuracy is 92.82%, the  stacking ensemble's is 93.94%, and the adaptive  boosting classifier's is  93.63%. From among the six  models that were taken into consideration for the  study, these findings demonstrate that the Stacking   Ensemble classifier proves to be the best-performing  model in terms of phishing detection. The models'  performances are acceptable  and might be used to  identify phishing assaults on e-commerce websites.