Overview
EN ISO/IEC 5259-2:2025 (adoption of ISO/IEC 5259-2:2024) - titled Artificial intelligence - Data quality for analytics and machine learning (ML) - Part 2: Data quality measures - defines a data quality model, a catalogue of data quality measures, and guidance on reporting data quality specifically for analytics and machine learning. Published by CEN in May 2025, this document is applicable to all types of organizations seeking to meet data quality objectives for AI, analytics, and ML projects.
Key topics and technical scope
The standard organizes data quality for ML and analytics into components, a model, and measurable characteristics. Key technical topics include:
- Data quality model and lifecycle components for analytics/ML tasks.
- Data quality characteristics and associated measures (definitions, scope, and guidance).
- Data quality reporting framework and required measure metadata for transparent reporting.
- Guidance for implementing measurement functions and integrating measures into workflows.
Representative data quality characteristics covered (each with measures and guidance) include:
- Accuracy, Completeness, Consistency, Credibility, Currentness
- Accessibility, Compliance, Efficiency, Precision, Traceability, Understandability
- Availability, Portability, Recoverability
- Auditability, Balance, Diversity, Effectiveness, Identifiability, Relevance, Representativeness, Similarity, Timeliness
The standard also provides informative annexes such as measurement-function design, a UML model of the measure framework, and a comparison with related quality models (e.g., ISO/IEC 25012).
Practical applications
EN ISO/IEC 5259-2:2025 is practical for:
- Assessing dataset readiness before model training (dataset validation, completeness, representativeness).
- Setting up continuous data-quality monitoring and drift detection in ML pipelines.
- Producing standardized data quality reports for governance, audits, and regulatory compliance.
- Defining acceptance criteria for data suppliers and datasets used in analytics or AI systems.
Use cases: dataset curation, model development QA, data governance frameworks, supplier data contracts, and compliance reporting.
Who should use this standard
- Data scientists and ML engineers
- Data stewards, data quality and governance teams
- Data architects and platform engineers
- Compliance officers, auditors, and risk managers
- Organizations deploying analytics/AI at scale seeking repeatable data-quality practices
Related standards
- ISO/IEC 5259 series (other parts addressing broader AI data quality)
- ISO/IEC 25012 (data quality model) - compared in Annex E of the standard
Keywords: EN ISO/IEC 5259-2:2025, ISO/IEC 5259-2:2024, data quality, machine learning, analytics, data quality measures, AI data governance.