Hi all.
I have a simple question for you 
We would need to create a percentile measure in an olap cube. As far as know, the measure is not available as a predefined aggregation, but...is there a way to obtain it, maybe via mdx or via information map? The problem is that this kind of measures, as a nunique, would rely on all the original dataset, and cannot be preaggregate...
I have tried by making a calculated mesure by making a key dimension for each row, and calculating the measure as
pctl5(90, SetToList({[KeyDimension].CurrentMember.Children},[Measures].[MeasureToUse])),
where measure to use use a mean aggregate function (but going to keyDimension I would have the mean of every single value, so the value itself).
but the response is:
Insufficient memory - in the "CHILDREN" function
Has anyone any feasible solution?
Thanks! 🙂