Glossary · IIoT, data and AI
Data quality
German: Datenqualität
In data management, data quality is the degree to which data is fit for its intended use, assessed by characteristics such as accuracy, completeness, consistency, timeliness, validity and uniqueness. ISO/IEC 25012 defines a data quality model; the ISO 8000 series addresses data quality for master data and data exchange.
- IIoT
- Standards
In one sentence
Data quality is the degree to which data is fit for its use, judged by accuracy, completeness, consistency, timeliness and validity.
Example
A temperature tag delivers values every second but with a frozen value for two hours after a sensor failure, which is timely yet inaccurate data.
How it applies
- Engineering: Quality problems in industrial data include wrong units, frozen values, time stamp offsets, missing context and mislabeled events. Many can be caught by automated checks at the point of acquisition (Data validation).
- AI: Model quality cannot exceed the quality of its Training data; errors in labels or sensor data are learned as if they were facts.
- Documentation: Documentation depends on data quality too: product data, technical data tables and spare part numbers should be validated at the source. State known data limitations in reports and model documentation.
Data quality vs. data integrity
Data quality asks whether data is fit for use. Data integrity asks whether data is complete, consistent and protected from unauthorized or accidental change over its lifecycle (see Sensor data integrity). Data can be intact and still of poor quality, for example if the sensor was wrongly calibrated.