NVD CVE Dataset: Cleaned Vulnerability Records from 2004 to 2025

A cleaned and enriched dataset of Common Vulnerabilities and Exposures (CVE) records from the National Vulnerability Database (NVD) spanning 2004 to 2025. The data includes severity scores, CVSS metrics, CWE identifiers, and references, and is described as being prepared for machine learning tasks. The original source is the NVD, and it was aggregated and processed by an author on Kaggle.

Use Cases

• Predict vulnerability severity based on CVSS scores and CWE types mentioned in the description
• Classify vulnerability types using the CWE (Common Weakness Enumeration) identifiers
• Analyze temporal trends in disclosed vulnerabilities across the 2004-2025 range
• Build a knowledge graph linking vulnerabilities to external references and advisories

Strengths

• Covers a 22-year time range from 2004 to 2025, providing longitudinal data
• Includes multiple structured fields such as severity, CVSS, and CWE for analysis
• Data is described as cleaned and enriched, suggesting preprocessing for usability

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment for large-scale ML

Provenance

Source

National Vulnerability Database (NVD)

Collection Method

Aggregated, cleaned, and enriched from the NVD source

Time Range

2004 to 2025

Freshness

Last update date is unknown; freshness unverified

Geography

Global