Machine Learning based Phishing Website Detection

Phishing websites are those where a person can imitate a legitimate website and trick an innocent user into thinking he is visiting a legitimate website while stealing sensitive information. Statistics from the Anti-Phishing Website Group (APWG) show that in the second quarter of 2023, 1,286,208 phishing attacks were recorded. This was the APWG's third-highest quarterly total ever. To address this issue, a Phishing website detection system was made that automatically detects and notifies the user of any phishing website. The system will be based on supervised learning wherein machine learning models such as k-nearest Neighbors, Logistic Regression, Support Vector Classifier, Naive Bayes, Decision Tree, and Random Forest are selected for training. Random Forest was found to have the highest accuracy among them. As a result, Randomforest was chosen with an accuracy of 96.7 and a combination of 32 features, among which features like status bar customizing, website forwarding, stats report, pop windows, server form handler were uniquely identified and implemented to the proposed system to improve the overall accuracy of the model.