I have a PBI data model, which is a star schema, but am wondering if it's optimal or 'correct' when it comes to creating DAX measures.
My fact table in PBI looks like this; it's a database view:
development_PK development_spend
DEV0001 £1000
DEV0002 £3000
DEV0003 £2000
To bring in a new dimension (something I want to filter or slice by), I create a view in the database that has the development_PK and the relevant value from the location table:
development_PK location
DEV0001 UK
DEV0002 France
I'll then join on development_PK in the PBI data model.
I could, of course, bring the primary key from the location table into the view I'm using for the fact table, where it would be a foreign key, and then simply connect to the location table.
My question is abstract in the sense that I'll be creating more data models and want to 'future-proof' them - I know certain ways of data model design can cause undesired results when writing DAX measures, e.g. in a single table model, auto-exist will create unexpected row counts with certain slicer configurations.