Phishing Detection System through Hybrid Machine Learning Based on URL

Abstract

Phishing attacks on the internet using a comprehensive dataset based on phishing URLs. The study utilizes various ML approaches like DT, LR, RF, NB, GBC, SVC and an innovative hybrid LSD model for enhancing cyber threat detection. In continuation, we have used a hybrid approach of hybrid multiple models, among which Stacking Classifier, an ensemble learning technique, has been used to merge the RF Classifier and MLP Classifier (as base classifiers). It is a LGBM Classifier based meta-estimator for the final prediction and therefore the extendability of the project to improve classification performance is enhanced. The effectiveness of the model is evaluated using evaluation metrics such as precision, accuracy, recall and F1-score. Based on the results, the hybrid LSD model is effective in combating phishing attacks and can provide a complete security solution for new cyber threats. The findings of this research can assist in strengthening cybersecurity measures and demonstrate the potential of ML in boosting internet security.