Phishing URL Model Added
Description: Phishing detection using ML models is crucial due to their ability to analyze patterns and features in URLs, enabling the identification of suspicious links, adaptation to evolving phishing tactics, and reduction of false positives, thus providing robust defense against phishing attacks in today's digital landscape.
Implementation:
A Flask application has been developed to deploy a Gradient Boosting Classifier trained on a dataset of over 11,000 website URLs, each with 30 parameters and corresponding phishing labels. The dataset was split into 80-20 train-test sets, and after training various SVM models, the Gradient Boosting Classifier emerged as the best performer, even amidst varying noise levels. The Flask app, housed in app.py, offers a POST request endpoint for predictions, with dependencies detailed in requirements.txt, while the training data resides in phishing.csv. This setup enables efficient deployment of the ML model for identifying phishing websites.
Current Output: