URL based Phishing Detection System using Ensemble of LSD Model

Phishing is a deceptive process on the Internet which involves forming of fake web pages to trick people into sending confidential information to the phisher. These websites are made to appear real in order to trick people. The goal of this research is to enhance the detection of phishing websites by classifying our websites using a variety of methods. Phishing uses email fraud as the foundation for misleading emails, then uses fake sites to get the necessary information from the target audience. Education is essential in combating cybercrimes related to phishing attacks. Various research has reported on the prevention, detection, and understanding of phishing harms; however, a comprehensive and effective method to mitigate them is currently lacking. To combat this, the suggested solution involves based a dataset with phishing URLs that was received from a reliable Kaggle source, collected from 11,000 websites in vectorized form. Well-known machine learning algorithms include logistic regressions (LR), support vector machines (SVM), decision trees (DT), and an ensemble hybrid model that combines grid search, soft voting, and canopy feature selection is used for improving the phishing detection in the system. Additionally, to evaluate the approach different parameters like precision, accuracy, recall, F1-score to illustrate the efficiency of the model. The accuracy yielded as 97.9 percentage and demonstrates that the proposed approach is more efficient than other existing models.