Website phising detection application using Support Vector Machine (SVM)

Phishing is an act to get someone's important information in the form of usernames, passwords and other sensitive information by providing fake websitest at are similar to the original. Phishing (fishing for important information) is a form of criminal act that intends to obtain confidential information from someone, such as usernames, passwords and credit cards, by impersonating a trusted person or business in an official electronic communication, such as electronic mail or instant messages. Along with the development of the use of electronic mediawhich is followed by the increase in cyber crime, such as this phishing attack. Therefore, to minimize phishing attacks, a system is needed that can detect these attacks. MachineLearning is one method that can be used to create a systemthat can detect phishing. The data used in this research is11055 website data, which is divided into two classes, namely "legitimate" and "phishing". This data is then divided using10-fold cross validation. While the algorithm used is the Support Vector Machine (SVM) algorithm which iscompared with the decision tree and k-nearest neighbor algorithms by optimizing the parameters for each algorithm. From the test results in this study, the best system accuracy was 85.71% using SVM kernel polynomial with values ofdegree 9 and C 2.5.

Keywords: Feature extraction, Machine learning, Parameter optimization, Phishing support vector machine(SVM).