Skip to content

0541 Data Grains

Data Grains define the granularity of data.

Data Grains were invented by the Kimball Group to describe the granularity of data in a data warehouse fact table. The concept fits well into a metadata landscape too. For example, you may have multiple data sources that capture sales data for your organization. One data source may contain daily figures, another contains weekly figures, and another contains monthly figures. They are all the same type of data if you consider there subject area. They may even have the same schema so they will be technically easy to combine. However, their different granularity means they can not be combined without transformation, and only at the most course-grained level, if distortion is to be avoided.

Data Grains are also useful when defining the data requirements for a project. They can be associated with a data field definition or associated with a data lens.

UML

DataGrain entity

DataGrainDefinition relationship

DataGrainAssignment relationship

DataGrainHierarchy relationship

Raise an issue or comment below