Forum Discussion
Aggregation on rows with user selected filters
Super UserIf you unpivot the features and define a custom column that groups the versions, this should be doable.
See attached.
I see your approach but the value of the feature for offer type 2 should be included when grouping. For T_MAX, version 6 changes from 150 to 150000 when comparing INC with INP. But the report shows two rows for 6 (empty -> 150000 and 150 to empty)
I also changed the value for 12 and the same:
As far as I can see, the value for offer type 2 should be included for grouping. And this means, combinations of types should be created as well. And this leads us to the original approach.
Am I correct?
- AlexisOlson
You'll have two rows regardless (either "6" and "9" or else "6" and "6,9"). The advantage of the way I did it is that the version strings can be pre-calculated. If you can come up with a better approach, then go for it.
Yes, I am afraid I should go for the initial approach and get the combinations, and the grouping for each feature. What I really wanted to know is that there is no possibility of doing it with just measures as it is done with Qlik. It's so easy; you just create a measure that returns multiple values and get the aggregation level that you need. You don't need to "pre-cook" all the data.
I guess there are other scenarios where PowerBI is easier than Qlik. Many thanks for your help and your time. Really appreciated.
- AlexisOlson
The difficulty here is that you are essentially wanting to treat the list of versions as both a dynamic measure and a dimension at the same time. This is a pretty unusual situation.
