
1. Re: Dividing a dataset into 4 categories that always appear
Justin Larson Oct 2, 2017 12:57 PM (in response to ann.stolzman)Great workbook, and good problem.
If you turn your Weighted Ratio into a FIXED LOD calculation, then it can be treated as a rowlevel calculation.
[Weighted Ratio per Dim]:
{ FIXED [Selected Dimension]:SUM([Quantity]*[Selected Measure])/SUM([Selected Measure])}
Then you can do your Rank Category calculation based on that as a row calculation as well, rather than a window calculation that relies on the LOD being on the canvas:
if [Weighted Ratio per Dim] > { FIXED [Select Dimension]: PERCENTILE([Weighted Ratio per Dim],.75)} then 'Tier 1'
elseif [Weighted Ratio per Dim] > { FIXED [Select Dimension]: PERCENTILE([Weighted Ratio per Dim],.5)} then 'Tier 2'
elseif [Weighted Ratio per Dim] > { FIXED [Select Dimension]: PERCENTILE([Weighted Ratio per Dim],.25)} then 'Tier 3'
else 'Tier 4'
end
This allows you to mark each row's Tier, but based on the aggregate across rows. You may have to alter the way you were using WeightedRatio  such as taking an average of it, rather than a sum, since now it's returning values per [Selected Dimension], then doing the selected aggregate.

2. Re: Dividing a dataset into 4 categories that always appear
ann.stolzman Oct 2, 2017 3:06 PM (in response to Justin Larson)Actually, it seems to work correctly in the model that I have with the real data. I knew it had to be something simple  I should have started with the LOD in the first place. I was approaching the LOD idea from the last mile of the problem instead.
Thanks for the solution, and for the compliment on the workbook.
Ann