Temporal Analysis and Common Weakness Enumeration (CWE) Code Prediction for Software Vulnerabilities Using Machine Learning

Zero-day attacks present a significant challenge in cybersecurity by exploiting vulnerabilities that have not been identified yet. This paper leverages the Common Weakness Enumeration (CWE) system—a widely recognized framework for identifying and categorizing software weaknesses—to predict potential zero-day attacks through temporal analysis. By examining historical CWE data, this study aims to uncover patterns that could indicate emerging vulnerabilities, thereby enabling the implementation of proactive security measures. This research offers a new method of predicting zero-day attacks through a systematic analysis of temporal trends, contributing to developing more complex cybersecurity defences. The work is unique and not previously documented in the literature, highlighting the importance of temporal analysis in forecasting and mitigating cybersecurity threats. The proposed predictive model, which employs statistical methods, time series analysis, and prediction model, achieved an accuracy of 0.66, a precision of 0.65, an F1 score of 0.64, and a recall of 0.66. These results demonstrate the model's balanced performance in effectively identifying relevant CWE codes while minimizing false positives.