Dimension and Measure

In Tableau, each field has a dimension or measure role. Dimensions are what you are analyzing; measures are how much of it there is. Tableau’s definitions: “Dimensions group or segment the data. Dimensions typically contain qualitative values (such as names, dates, or geographical data)”, while “Measures are aggregated. Measures typically contain numeric, quantitative values that you can measure.”

Country, region, product category, and month are dimensions — they carve the data into groups. Revenue, emissions, and response time are measures — they are summed or averaged within whatever groups the dimensions define. Put a dimension and a measure together and you have a summary; that is most of what a BI tool does.

In Tableau, a second distinction rides alongside, and the color of a field signals it. “Discrete means ‘individually separate and distinct.’ These fields are colored blue” and produce headers. “Continuous means ‘forming an unbroken whole, without interruption.’ These fields are colored green” and produce axes. Knowing which is which predicts what dropping a field into a view will do before you drop it.

The roles are assigned for you on connection: “Each field is automatically assigned a data type such as integer, string, or date, and a role: a discrete dimension or continuous measure.” Convenient, and the source of a common failure — a numeric column that arrived as text is classified as a dimension and will not aggregate, so totals silently go missing.

The habit worth forming is to read the field list before building anything: check that every quantity you intend to sum is a measure, and that identifiers you never want summed — an account number, a ZIP code — are not. Correct the type at the data source rather than working around it in each view.

References:

• Data types and roles