Phishing website analysis and detection using Machine Learning

Abstract

Cybersecurity has become an essential part of this new digital age with more than 820 million users of internet by the year 2022 there is need of security systems to protect public from phishing scams as it not only effects the wealth but also effects the mental health of public, making people afraid to surf or use the internet services which motivates me to work on this problem to develop efficient solution. Objective of this paper is to analyse some common attributes shown by phishing websites and develop a model to detect these websites. Various models were trained on the dataset like Random Forest Classifier, Decision Tree Classifier, Logistic Regression, K Nearest Neighbours, Artificial Neural Networks and Max Vote Classifier of Random Forest, Artificial Neural Networks and K Nearest Neighbours. Highest accuracy was achieved by Max Vote Classifier of Random Forest (max depth 16), Decision Tree (max depth 18) and Artificial Neural Network of 97.73%. This research can be used in real life by implementing a web application in which user can enter the website link and using the link the application will get values for various factor on which model was trained and it will detect whether a website is phishing website or not.