Overview
EN ISO/IEC 5259-3:2025 (ISO/IEC 5259-3:2024) - Artificial intelligence - Data quality for analytics and machine learning (ML) - Part 3 - specifies requirements and guidance for establishing, implementing, maintaining and continually improving the quality of data used for analytics and ML. The standard defines a data quality management framework and a reference set of processes and methods that organizations can tailor. It is intentionally generic and does not mandate specific metrics, detailed processes or tools, making it applicable across organization types and sizes.
Key topics and requirements
- Overall data quality management: objectives, governance, data quality culture, handling data quality issues, competence and resource management, documentation, audits and confirmation reviews.
- Life cycle-specific management: guidance across the data lifecycle including
- Data motivation & conceptualization
- Data specification and planning
- Data acquisition
- Data preprocessing and augmentation
- Data provisioning and decommissioning
- Horizontal processes: verification & validation, configuration management, change management and risk management applied to data used in ML/analytics.
- Supply chain considerations: managing data quality across suppliers, third-party data providers and partnerships.
- Management of data processing tools: requirements and recommendations for tools that collect, process or transform data.
- Work products: lists of outputs (e.g., specifications, plans, audit records) tied to lifecycle stages to support implementation and compliance.
Note: the standard defines requirements and guidance, not prescriptive metrics or fixed methodologies. Organizations are expected to tailor processes to meet these requirements.
Applications and who should use it
EN ISO/IEC 5259-3 is relevant to:
- Data engineers, ML engineers, data scientists and data stewards designing and operating ML pipelines.
- AI governance, quality assurance and compliance teams establishing data governance, auditability and lifecycle controls.
- Procurement and vendor managers who need to specify data quality expectations in supplier contracts.
- Tool vendors and platform providers aligning product capabilities with recognized data quality management requirements.
Practical uses include building data governance programs, defining data quality controls for ML training/validation, documenting work products for audits, and reducing risks from poor data in analytics systems.
Related standards
- Part of the ISO/IEC 5259 series for AI data quality and aligns with ISO/IEC JTC 1 “Information technology” AI guidance. Adopted as EN ISO/IEC 5259-3:2025 by CEN for use across European national standards bodies.
Keywords: data quality, machine learning, analytics, data quality management, ISO/IEC 5259-3, EN ISO/IEC 5259-3:2025, AI data governance, data lifecycle, risk management.