Phishing Website Classification: A Machine Learning Approach

In order to circumvent the adverse effect of fraudulent acts committed on the internet by adversaries, different researchers have proposed various solution to this problem. One of this online fraudulent act is website phishing. Website phishing is the act of luring unsuspecting online users into divulging private and confidential information which can be used by the phisher in fraud, blackmail or other ways to negatively affect the users involved. Based on our previous paper, we proposed noble features to better improve the accuracy of machine learning algorithms in classifying phish. In order to ascertain the improvement in website phish classification of machine learning algorithms based on the features extracted in our previous paper, our present approach is based on testing. This approach is divided into three phases. In phase 1, we propose a new method of classifying phish website by using pruning decision tree. In phase 2, we train and test four selected individual reference classifiers and based on their performance, an ensemble of classifier is designed. Lastly, the output of each phase is then compared to show the efficiency of our approach. The experimental result of the research shows that pruning decision tree is comparatively potent in website phish detection.