Dealing with more than one Fact Table using “Linked Table”

Question

Create a model with more than one Fact Table is a common scenario, and usually we have to join these Fact Tables against a common set of dimensions.Common problems of this scenario are circular references (Indirect relations that prevent to activate a direct relation) and models difficult to read and maintain.I would like to share a practice known as “Linked Table” acquired in my past experiences on different BI tools (QlikView, Cognos).A “Linked Table” can be expressed in PowerBI as a table that contains the common fields from two or more tables.In a recent project, I need to create a model in PowerBI to analyze Purchase and Payments Statistics of a Retail Store Chain.The below table contains a simplified scenario in order to focus on the “Linked Table”:*Be aware that Payment Statistics are not available on Product Category.Just for highlight circular references in this simply scenario I enabled Cross Filtering Dimension Option on Both.As you can see, model prevent to create a relation between Vendor and Purchase Order since this is already filtered by indirect relations:Fig 1: Data Model with Circular ReferencesIf you try to enable the inactive relation you get the classical error in fig2:&nbsp;Fig 2: Error trying to enable the inactive relation&nbsp;In order to allow the direct slicer from Vendor Dimension to the Purchase Order we can enhance the model adding the linked table following these steps:Create a Linked Table containing an Identity (LinkedTableKey) for each couple {Company Code; Vendor Code} as set from the Fact Table (Purchase Order and Payment Statistics);Replace the fields {Company Code; Vendor Code} in the Fact Table (Purchase Order and Payment Statistics) with the related field LinkedTableKey created in the step 1;Link Vendor and Company directly to the LinkedTable.&nbsp;Fig 3: LinkedTable Model&nbsp;As you can see in Fig3, LinkedTable approach allows your Model to slice both Fact Tables, in addition, the model is more readable and maintainable.Of course, this is just one of many approach to solve the Scenario.&nbsp;How did you deal similar Scenario? Do you plan to use Linked Table in your projects?

anonymous · Answer

hi&nbsp;gipiluso,&nbsp;Thanks for the solution. But I have a problem. I have 3 facts and 4 dim tables:&nbsp;F1 - links to D1,D2,D3F2 - links to D2,D3,D4F3 - links to D3,D4&nbsp;In this scenario, how do I set up the linked table ? Do I just create 1 broad linked table which (with columns linking to all Dimension tables) and has got distinct entries from each fact table, and NULLs where the dimensions are absent ?&nbsp;&nbsp;thanks,&nbsp;

anonymous · Answer

Hi&nbsp;gipiluso,
&nbsp;
Thank for sharing, I think these will be help for others who has the similar scenario.
&nbsp;
Regards,
Xiaoxin Sheng

gipiluso · Answer

If you can create just one linked table i suggest to go for it

Forum Discussion

Related content

Dataverse Virtual table link to Fabric

Dataverse – Fabric Link: Metadata Tables No Longer Synced

Fabric Dataverse - Synapse Link versus Fabric Link

Dealing with multiple fact tables?

Dataverse tables enabling themselves on Link to Fabric