PERBANDINGAN KLASIFIKASI ALGORITMA K-NN, NEURAL NETWORK, NAÏVE BAYES, C 4.5 UNTUK MENDETEKSI WEB PHISING

Abstract

The increasing number of internet users in the world and the rise of web phishing. Based on the Wearesocial report, there are several facts which include the number of world internet users that have reached 4.021 billion people. explained that of the hundreds of millions of internet users in Indonesia, 60 percent have accessed the internet using smartphones. Internet usage is dominated by socializing activities in cyberspace. Evidenced by the large number of world social media users, reaching 3,196 billion users. But with the rise of internet users and social media accompanied by the proliferation of phishing webs. The purpose of this attack is to make users believe that they interact with the official virtual or online site in question. Generally the information sought is in the form of a password, account or victim's credit card number. by way of being directed to a fake web or sending an email, banner or pop-up where the victim is asked to give his personal information. In this study C.4.5, Neural Network and K-Nerest Neighbor algorithm comparison will be conducted to determine the performance of the algorithm in detecting webphising by using classification techniques. Based on the model testing based on the Naïve Bayes method, Decision Tree C.4.5, K-NN, Neural Network uses the Weka framework v.3.8.2. The results of the Decision Tree C.4.5 algorithm have a higher level of accuracy with 89.66 percent accuracy.