Identity Theft and User Cybersecurity Behaviors: A Machine Learning Framework for Predicting and Mitigating Phishing and Social Engineering Attacks Risks
This research presents a machine learning framework designed to predict and reduce the risk of identity theft caused by phishing and social engineering. The authors developed a Cyber Risk Score (CRS) that combines observable security habits, like password hygiene, with latent psychological traits such as impulsive link-clicking. By utilizing a hybrid stacking ensemble model, the study achieved a 93% accuracy rate in identifying vulnerable social media users. Beyond mere prediction, the system uses SHAP analysis to provide transparent, personalized recommendations tailored to an individual’s specific behavioral weaknesses. This user-centered approach aims to bridge the gap between cybersecurity knowledge and actual online behavior through evidence-based interventions. Ultimately, the framework offers a scalable, ethical solution for organizations to protect users in increasingly sophisticated digital environments.