Predicting Individuals’ Vulnerability to Social Engineering Attacks

Abstract

The increasing prevalence of social engineering attacks (SEAs) poses a significant threat to individuals and organizations [1]. This study aims to develop and evaluate a predictive model for identifying individuals susceptible to SEAs. A dataset comprising 505 responses from university and colleges students in Addis Ababa, Ethiopia, was collected through an online survey. To address data insufficiency, synthetic data generation was employed, expanding the dataset to 3000 instances [2]. Employing a Design Science Research (DSR) methodology, multiple machine learning algorithms were compared, with Random Forest demonstrating superior performance in predicting SEA vulnerability. Key predictors identified include age, income, occupation, curiosity, and trust factors. The findings contribute to enhancing cybersecurity measures by enabling targeted interventions to protect vulnerable individuals.