An Optimized Bagging Learning with Ensemble Feature Selection Method for URL Phishing Detection

This study proposes and implements an ensemble feature selection with a bagging classifier for URL phishing detection. Feature Selection is essential for reducing data's dimensionality and improving any proposed framework's performance. The feature selection stability is improved by using the ensemble feature selection method. In this work, Aggregation decides the final ranking of the ensemble feature selection by using four standard filter methods. Bagging classifier used for URL phishing dataset and accuracy of the model is determined with aggregation ranked features. In proposed work details the ensemble feature selection methods that embed with optimized ensemble bagging learning. The hyperparameter of the bagging classifier, such as multiple estimators with random patches, random subspaces, bagging, or bootstrap aggregation and pasting, are tuned, which produces the better performance model. The evaluation and comparison of experimental results showed the effectiveness of our method.