An Ensemble Learning Model for Predicting Social Engineering Pharming Attacks

Abstract

Pharming attack has a broad scope as social engineers can masquerade as anyone, particularly during the COVID-19 pandemic from health authorities or even organization executives getting in touch with their personnel. The study aims to develop an ensemble model for predicting social engineering-based pharming attacks from the client-side pharming attack. The target population for the study includes 1781 URLs, which are secondary and readily available on Kaggle having been compiled by Manu Siddhartha. The study focused on identifying URLs that facilitate Pharming attacks, a cybersecurity threat. Malicious URLs miss the protocol segmentation, certificate of authorization, and bait targets into becoming victims of pharming attacks. The research instrument used for data process and building a pharming attack model was CSV and Jupyter notebook. Data was collected from secondary data sources. A model for predictive pharming attacks was built utilizing Logic regression, Random Forest, and gradient boost as the model for boosting algorithms to reduce pharming malware attacks.