An Effective Approach for Phishing Detection Using Machine Learning Algorithm

Abstract

The trend of tricking web users has kept pace with the expanding utilization of online surfing. The rise of phishing attacks postures a noteworthy risk to individuals and organizations everywhere. Phishing is constantly advancing to receive modern methods and techniques to steal important pieces of information from users. Phishing is a form of attack initiated by an email or social media message which mainly forwards the casualties to malicious web pages and these are extremely difficult to identify for security administrators. Phishing is a part of social engineering. Through this, hackers design a web page duplicate and send it to the user when the user enters information that data is directly saved to a database created by hackers. The most commonly used phishing techniques are link manipulation, filter evasion, website forgery, social engineering, and covert redirect. To recognize unique patterns, Machine Learning algorithms continuously learn from huge bulk data and in most research, it has been claimed that machine learning-based methods are more effective than other methods. Here, we use Five machine-learning classification techniques to detect phishing web pages and legitimate web pages with desirable accuracy. In our work, we apply Logistic regression, Decision tree, XGBoost, Random Forest, and SVM algorithms. All algorithms perform incredibly well on dataset. The Random Forest algorithm surpasses them all with a 98% accuracy rate.