Forum Discussion
Modelling Problem
Hi Cling ,
Here I create a sample to have a test. I think by default the direction for one to many relationship is single.
I suggest you to keep single instead of choose both between A and INTERMEDIATE / B and INTERMEDIATE.
In my sample the table based on [DESC_A] and [DESC_B] is the same as yours.
Data model:
Then if I click the data which only from Table A or Table INTERMEDIATE, the data from other table won't be filtered.
However if I click the table contains data from Table B, it will be filtered. This is because that [DESC_B] from B table could interact data from other table.
I suggest you to turn off the interaction in visuals which you don't want it be impacted by others.
For reference: Change how visuals interact in a report - Power BI | Microsoft Learn
Best Regards,
Rico ZhouIf this post helps, then please consider Accept it as the solution to help the other members find it more quickly.
Super UserHi Cling - you want to establish a bidirectional relationship between Dimension A and Dimension B through an intermediate table without affecting other fact tables.
as per the above images, Dimension A to Intermediate Table: Create a one-to-many relationship (ID_A).Dimension B to Intermediate Table: Create a one-to-many relationship (ID_B).Both relationships should be bidirectional to allow filtering in both directions.
By default, Power BI will allow cross-filtering between Dimension A and Dimension B for all related tables, which can affect other fact tables.
you can prevent below type:
Identify the relationships between Dimension A, Dimension B, and the other fact tables.
Change these relationships to single direction (from the dimension to the fact table).
Create a custom table:
Use the intermediate table to display the relationships between Dimension A and Dimension B without propagating filters to other tables:
Add DESC_A from Dimension A and DESC_B from Dimension B to a table visual.
Power BI will use the intermediate table to link the two dimensions, showing the correct results.
Doing what you indicated, the filter would propagate, If I filter in A, the filter would propagate to B through the intermediate table and affect the other tables: