Viewing the data model

The Semantic Layer page shows the model in an editor, with a side panel that describes what is saved. To open it, go to Data Catalog and click Semantic model.

The page at a glance

• The top bar has a back arrow to the Data Catalog, an Unsaved badge when you have edits that are not saved, and the Find joins, Check, and Save buttons.
• The editor shows the model in Build or Code. See Editing fields.
• The side panel has four tabs: Diagram, Used by, Suggested joins, and Syntax guide. On smaller screens it appears below the editor.

Diagram and Used by show the saved model. Edits you have not saved appear there after you save.

Before a model is saved

Until you save a model, the editor opens on a starting draft. If the workspace already has metrics, the draft is written from them, with one business object for each table they use. Otherwise it is a commented template. A notice above the editor reminds you that nothing is saved until you choose Save.

Diagram

Diagram draws your saved business objects and the joins between them, with a summary of how many objects, sources, and joins the model has. Below the drawing, objects are listed by database, each with its table and the joins it starts.

• A green check marks a certified object.
• no dimensions flags an object whose metrics cannot be split yet.
• cross source marks a join that reaches into another database. These joins work, but filter or summarise large tables before joining them.

When nothing is saved yet, the tab shows No model saved yet.

Used by

Used by traces each business object to the metrics built on it, and each metric to the dashboard cards and scheduled tasks that read it. Use it to see what a change will affect before you save. A metric that nothing reads shows Nothing reads this yet.

Metrics saved directly against a table appear in a separate group, Outside this model. Saving the model does not change them.

Suggested joins

Click Find joins to look for joins between your saved business objects. Qubit proposes candidates from column names, such as a store_id column next to a stores object, then samples real rows to see whether the values match. Each suggestion is labelled:

• Confirmed in data: the sampled values were found in the other object.
• Partial match: only some of them were. Review these before you use them.
• Cross source: the two objects are in different databases.

Click Add to spec to insert a suggestion under the right business object in the editor, or Copy to copy its YAML. Suggestions are not saved until you save the model.

Syntax guide

Syntax guide explains each section of the YAML document and its fields, with examples you can copy.