URL Phishing Detection using Machine Learning Technique

One of the primary worries of security researchers nowadays is the staggering number of phishing attempts. Traditional phishing website detection technologies rely on signature-based techniques that are incapable of detecting recently generated phishing websites. As a result, researchers are developing Machine Learning-based algorithms capable of detecting and classifying phishing websites with high degree of accuracy when a vast number of characteristics are evaluated. Building a classification model with a vast number of characteristics, on the other hand, requires time, which impedes the rapid recognition of phishing websites. As a result, it is important to use a feature selection approach to shortlist a collection of features so that high-performance classification models may be constructed in less time. The performance of Machine Learning methods with and without feature selection is compared. Experiments are carried out on a phishing dataset with 30 characteristics, which includes 4898 phishing and 6157 legitimate websites. According to the comparison findings of the applied classification algorithms, the Random Forest(RF) algorithm performs the best at detecting phishing URLs, with a 91.19 percent accuracy rate