Research on Global Cybercrime Prediction Based on Regression Analysis and Machine Learning

Abstract

This article establishes a comprehensive model for evaluating cybersecurity policies, revealing the global distribution pattern of cybercrime through data-driven analysis, and exploring which national policies have performed well in preventing, combating, and prosecuting cybercrime. By analyzing the global Cybersecurity Exposure Index (CEI) and the heat map of cybercrime distribution, it reveals that some countries in Europe, Asia, and the Americas face higher cybercrime risks due to advanced information technology and frequent online activities, while some countries in Africa and Oceania have relatively fewer cybercrimes due to lower digitalization levels. In addition, through the gray correlation analysis (GRA) model, the study found that political stability, economic development and cultural factors are significantly associated with cybercrime. This article develops a network security policy evaluation model. The decision tree regression model is used to evaluate the impact of different network security policies. The results indicate that organizational measures have the highest characteristic importance (66.60%), followed by technical measures (18.00%). Although legal measures are important, their impact is relatively small (4.00%). This indicates that effective organizational strategies and technological implementation are crucial for reducing cybercrime.