I had seen this suggestion (manually creating your Calendar table for Time Intelligence), but didn't think it was important. Until now.
After creating a data model that auto-generated the hidden date table, it was time to run my measures through some validations. In testing a measure that looked as if it was computing results correctly, I found that indeed it wasn't. Several of the values that I expected in the result set were missing. After spending time in Dax Studio (lots of head scratching, rewriting the measure, etc.),the problem surfaced - it had to do with date values being ignored/removed/context-transitioned-out? of the result set. If my expected result set all had a single date value, the expected result was returned. If my expected result set contained multiple dates, only a subset was returned and all of that data corresponded to a single date. Note: part of the visual filtering criteria included "Dates Between" on a slicer.
After trying various solutions, the only one that worked definitively was to follow the advice of Russo/Ferrari - manually create a CalendarDate table and mark it as the Date Table. Magic happened and PowerBI started to give consistent & complete results as expected.
I apologize if this is a repetitive post, but I haven't been able to find other sources where people experienced the same situation as I have.
Just happy it's working 🙂
