ISO IEC 5259-2-2024 PDF

Full title and description

ISO/IEC 5259-2:2024 — Artificial intelligence — Data quality for analytics and machine learning (ML) — Part 2: Data quality measures. This part of the ISO/IEC 5259 series defines a data quality model, a set of measurable data-quality characteristics and guidance on reporting data quality specifically for analytics and machine learning workflows.

Abstract

ISO/IEC 5259-2:2024 specifies a data-quality model and practical measures that organisations can use to assess, quantify and report the quality of data used in analytics and ML systems. It aligns with existing data-quality work (for example ISO/IEC 25012 and ISO 8000), covers both structured and unstructured data, and provides guidance to support transparent, repeatable data-quality assessment across the data lifecycle.

General information

• Status: Published.
• Publication date: November 2024 (published; national catalogues report 5 November 2024).
• Publisher: Joint ISO/IEC publication (ISO and IEC, developed by ISO/IEC JTC 1/SC 42).
• ICS / categories: 35.020 — Information technology (IT) in general / Artificial intelligence and data for ML.
• Edition / version: Edition 1 (2024).
• Number of pages: 38 pages (English edition reported as 38 pages by multiple national distributors).

Scope

This document specifies a data-quality model, measurable data-quality characteristics and guidance on how to report data quality in the context of analytics and machine learning. It is applicable to organisations of any size that develop or use data for analytics/ML and covers both structured (tables, records) and unstructured data (text, images, audio) commonly used in ML pipelines. The standard is intended to be used alongside ISO/IEC 5259-1 (framework and terminology) and other data-quality standards.

Key topics and requirements

• Definition of a data-quality model tailored to analytics and ML, including relationships between quality characteristics.
• A catalogue of measurable data-quality characteristics (for example: completeness, accuracy, consistency, validity, timeliness, uniqueness, provenance/lineage and fitness-for-purpose) and guidance on how to measure them for different data types.
• Guidance on reporting and communicating data-quality measurements to support transparency for ML development, validation and governance.
• Alignment and mapping to existing data-quality standards and metadata practices to support interoperability (references to ISO/IEC 25012, ISO 8000 and related work).
• Applicability to both structured and unstructured data used across the data lifecycle (collection, curation, labeling, training and inference).

Typical use and users

Primary users include data engineers, data stewards, ML engineers, data scientists, compliance and governance teams, auditors and architects who need standardised ways to quantify and report data quality for analytics and ML projects. Organisations adopt the standard to improve model reliability, reduce bias attributable to poor data, and to support regulatory and procurement requirements that demand documented data-quality practices.

Related standards

ISO/IEC 5259-2:2024 is part of the ISO/IEC 5259 series. It is intended to be used with ISO/IEC 5259-1 (framework and terminology) and other parts of the 5259 family. It also complements existing data-quality and data-governance standards such as ISO/IEC 25012 (data quality model) and ISO 8000 (data quality), and sits alongside broader AI standards produced by ISO/IEC JTC 1/SC 42.

Keywords

data quality, data-quality measures, data governance, analytics, machine learning, ML, data model, completeness, accuracy, consistency, timeliness, provenance, ISO/IEC 5259, JTC 1/SC 42

FAQ

Q: What is this standard?

A: ISO/IEC 5259-2:2024 specifies a data-quality model and measurable characteristics for assessing and reporting the quality of data used in analytics and machine learning.

Q: What does it cover?

A: It covers definition of quality characteristics and measures, guidance for measuring and reporting those measures, and applicability to structured and unstructured data in ML/analytics workflows.

Q: Who typically uses it?

A: Data engineers, data stewards, ML engineers, data scientists, governance/compliance teams and organisations that develop or deploy analytics and ML systems.

Q: Is it current or superseded?

A: ISO/IEC 5259-2:2024 is an active, published International Standard (first edition, published November 2024). Organisations should check national standards bodies for local adoption or corrigenda.

Q: Is it part of a series?

A: Yes — it is Part 2 of the ISO/IEC 5259 series; Part 1 provides the framework and terminology and other parts address related aspects of data quality for AI/ML.

Q: What are the key keywords?

A: Data quality, data-quality measures, analytics, machine learning, completeness, accuracy, consistency, timeliness, provenance.