Cybercrime Analysis Based on Bayesian Model

Cybercrime has emerged as a pervasive global issue, impacting nations worldwide. In response to this growing threat, we aimed to develop a data-driven framework for national cybersecurity policy. More specifically, we have combined statistical knowledge, and employed several models, including Bayesian methods, and XG Boost to investigate the global distribution of cybercrime, explore correlations between demographics and the incidence of cybercrime. We collected cybercrime data from State Statistics Bureaux and VERIS Community Database (VCDB). To address the limitations in data and model selection, we enhanced data screening and preprocessing steps and optimized the model application. We focused on data in 2023, incorporating cybercrime success, reporting, and prosecution rates as well as national security indices (GCI, NCSI, CEI) and other socio-economic factors such as GDP, education levels, internet penetration, and IT infrastructure. We explored how demographic factors such as Internet access, wealth, and education levels correlate with cybercrime distribution. Using Generalized additive model, Bayesian Model, and Machine Learning Model, we discovered that countries with higher levels of wealth (GDP) tended to experience more cybercrime, which is likely due to more advanced digital infrastructure and higher rates of reporting.