SSAS - How to create a geographical dimension?

StackOverflow https://stackoverflow.com/questions/23288621

Question

I'm working on a SQL Server Analysis Services multidimensionnal cube. I want to create a geaographical dimension to allow user to explore the data regarding to a specific country, region, department, city, store.

Country > Region > Department > City > Store > Sale

So I created a new dimension by selecting the Sale table including all cascading dependencies. Then, I dragged items'name to create a hierachy as the screen capture bellow explains.

enter image description here

This is the diagram of all relationships between attributes. enter image description here

Finally, after processing I can browse the dimension. However, there is something that scaring me, I can't browse stores, there are only country, region, department and city nodes. So how to configure to display stores ?

enter image description here

Is there a way to display only branches containing stores because it's not relevant to display 36 000 cities for only 80 stores.

As you can seen, there are three warnings :

Avertissement   10  AttributeRelationship [DimStore].[Sale].[Id_] : Le nom spécifié pour la relation d'attribut diffère du nom de l'attribut associé.       0   0   
Avertissement   11  AttributeRelationship [DimStore].[Store].[Name_] : Le nom spécifié pour la relation d'attribut diffère du nom de l'attribut associé.        0   0   
Avertissement   12  Dimension [DimStore] : Éviter les hiérarchies d'attributs visibles pour les attributs utilisés comme niveaux dans les hiérarchies définies par l'utilisateur.       0   0

Any idea to solve that ?

Was it helpful?

Solution

I would flatten all those tables into one using a SQL view or ETL into a Dimension table. Then I would add that to the Data Source View, then painfully repoint each attribute at that. This gives you the opportunity to shape the data to suit SSAS.

I would also return the Attribute Relationships to their original state - all related directly to the Key attribute. There is little benefit in arranging them hierarchically and it relies on the input data being perfectly structured now and into the future.

OTHER TIPS

Use the view in database to restrict to cities with as store as well as flattening data as mike suggested.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top