Phishing Detection System through Hybrid Machine Learning Based on URL

The main focus of this study is on hacking efforts, which are a popular type of online crime. To trick people and get their personal information, phishing scams usually use fake emails and websites to trick them. There is still no foolproof way to stop, find, or raise knowledge about phishing attacks, even though there have been many studies on the subject. ML is very important for fighting online crimes like scams. A set of fake and real URLs from different sites was used for the study. After the data was preprocessed, several machine learning methods were used to protect users from fake URLs and make sure they were safe. DT, LR, RF, NB, GB, KNN, and SVM are some of the ML models used in this work to protect against hacking attacks. A mixed LSD model with both soft and hard votes is also added. This model combines logistic regression, decision trees, and support vector machines. Grid search hyperparameter optimization, cross-fold validation, and canopy feature selection are used in the innovative LSD model. Model performance was assessed using F1-score, sensitivity, memory, precision, and accuracy. Comparative study showed that the recommended method provided better outcomes.