Comparative performance of machine learning methods for classification on phishing attack detection

• Main Authors: Siti Noranisah, Wan Ahmad, Mohd Arfian, Ismail, Edi, Sutoyo, Shahreen, Kasim, Mohd Saberi, Mohamad
• Format: Article
• Language: English
• Published: The World Academy of Research in Science and Engineering 2020
• Subjects: QA76 Computer software; TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://umpir.ump.edu.my/id/eprint/30101/1/Comparative%20performance%20of%20machine%20learning%20methods%20for%20classification.pdf; http://umpir.ump.edu.my/id/eprint/30101/; https://doi.org/10.30534/ijatcse/2020/4991.52020; https://doi.org/10.30534/ijatcse/2020/4991.52020

The development of computer networks today has increased rapidly. This can be shown based on the trend of every computer user around the world, whereby they need to connect their computer to the Internet. This indicates that the use of Internet is very important, such as for the access to social media accounts, namely Instagram, Facebook, and Twitter. However, with this extensive use, the Internet does not necessarily have the ability to maintain account security in mobile phones or computers. With a low level of security in a network system, it will be convenient for scammers to hack a victim’s computer system and retrieve all important information of the victim for their benefit There are many methods that used by scammers to get the important information where phishing attack is the simplest and famous method to be used. Therefore, this study was conducted to develop an anti-phishing method to detect the phishing attack. Machine learning method was proposed as suitable to be used in detecting phishing attacks. In this paper, several machine learning methods were studied and applied in detecting phishing attack. Experiments of the machine learning methods were conducted to investigate which method performed better. Two benchmark datasets were used in the interest to access the ability of the methods in detecting the phishing attack. Then the results were obtained to show the performance of each methods on all dataset.