A weighted ensemble model for phishing website detection using random forest and deep neural network
Fabiha Khan, Md Mehedi Hasan, Krishna Das · 2023
Today, the number of people using the Internet to buy and sell goods and services is rapidly increasing. Website- based applications have become more popular in many business sectors because of their low operating costs and platform independence. User's private information may be stolen as a result of this growth and used maliciously. One method for stealing all of a user's information is phishing, which can send consumers to websites with dangerous content. It is a significant cybersecurity threat that can cause severe financial losses and reputational damage. To avoid falling victim to a phishing attack, it is essential to verify the URL before entering any sensitive information and to use anti-phishing tools and software to detect and prevent phishing attacks. This paper describes an approach to detect phishing websites by developing a weighted ensemble model using random forest (RF) and deep neural networks (DNN). By combining the predictions of these two models, we are achieving higher accuracy than any individual model. With 99% accuracy, the suggested model performs better than the individual models, while the random forest model and the DNN model alone achieve 98% and 73% accuracy, respectively. DNN and RF together can potentially capture different aspects of the data and lead to better predictions. DNNs are highly effective at detecting complex patterns in large datasets, while RF can identify the most important features for classification and handle noisy and missing data. By combining these algorithms, it is possible to leverage their strengths and improve overall accuracy and robustness of detection.