Create Calculated Dimensions/Measures across datasets

[Former Member]

Is there a way for Lumira to create a calculated measure or dimension using two different data sets?  For example, if I have Dimension 1 in Dataset A and Dimension 2 in Dataset B, can I create a calculated measure or dimension that divides Dimension 1 by Dimension 2?

[varunanand · Accepted solution]

I believe there will be some relation between these dimensions or measures.  So you would want to also evaluate between merge and blending.

SAP Lumira – Dataset Linking - YouTube

[Former Member · Accepted solution]

Unless I'm totally messing up the blending, this is not working.  I can create the measure using a blended data set, but when I try to drag it into a crosstab, it does not do anything.

[varunanand · Accepted solution]

Hi Jay,

Would you be able to give a little more information on the data set you are linking and the final visualizations with screen shots.

Thank you,

Varun Anand

[Former Member · Accepted solution]

I've been using a crosstab to validate my numbers.

I am trying to effectively combine different levels of detail in my calculations.  This is because of the fact that when Lumira connects to BEx, some of the calculations we do on our BW are coming across as "aggregation none".  There is apparently an issue in how Lumira processes aggregated measures in BW that causes this error, and it has been the source of a number of issues.

The basic issue I'm trying to overcome is that I need to create a SUM aggregation to do a simple division.  My numerator is coming from Datasource A while my denominator is coming from Datasource B.

• Datasource A contains multiple dimensions (including my primary key dimensions plus several additional dimensions).  The presence of these additional dimensions causes Lumira to calculate average values for every distinct combination of additional dimensions used in the data pull, regardless of whether or not they are included in a viz.  This causes my average value (which is basically a sum of all the averages for the distinct combinations)
• Datasource B contains only primary key dimensions (matching from Dataset A), without the additional dimensions.  This is so that when I calculate the sum of my average, I am minimizing the number of average calculations that go into my sum.  This gets me very close (if not spot on) to my true number.

The simplistic example I have above points to the problem - lumira calculates dataset B's total as the sum of the averaged values.  Because of this, I have to combine it with dataset A, which does the bare minimum number of average calcs before getting to the total, and then do my division.

Hopefully this made sense.

[TammyPowlas · Accepted solution]

Jay, on this:

Jay Chang wrote:



This is because of the fact that when Lumira connects to BEx, some of the calculations we do on our BW are coming across as "aggregation none".  .

Why not fix the aggregations back in the BEx query? Is that possible?  Then do what you need to do in Lumira

[Former Member · Accepted solution]

Because the aggregation cannot be fixed in the backend.  The way that particular measure is put together, it requires double aggregation.  This is not recognized by the Lumira to BEX import engine.

[varunanand · Accepted solution]

Hi Jay,

I'm trying to understand your problem based on the example. The data blending would complicate your calculation further and I would instead use custom calculations.

Quick question, why are the values not adding up to be the same in both the cases ? It just appears to be more detailed in dataset B. Can you check a couple more cases and see if the difference in the denominator can be overcome by simple round off ? In this case, we can round off the sum and use values from dataset B itself.

Thank you,

Varun Anand

[Former Member · Accepted solution]

The values do not equal each other across my two datasets on purpose.  This issue is the root of why I was going down this path.

Because of limitaitons in Lumira's processing of data from BEx, we have a situation where we cannot simply pull measures into Lumira from a BEx query.  Instead, I need to re-create those values on the Lumira side.  However, when I do that, then I run into my problem.

This is really difficult to explain without going into intricate details, but believe me when I say that the mis-match in totals across the two datasets in my example is deliberate.  What I'm trying to figure out is, given the two datasets shown, is there a way to calculate a measure that uses data from dataset 2 as the numerator and data from dataset 1 as a denominator?  Lumira lets me do this but then when I try to display that measure, Lumira won't let me drop it onto a viz.

[varunanand · Accepted solution]

Hi Jay,

If you are perform the linking of data sets correctly, you should be able to use both the measures and use the new formula measure in the visualization. Please look at this example below,

Please let me know if there are any questions or comments.

Thank you,

Varun Anand

[former_member696229]

Hi Fellows

I'm trying set up in Material Master the Denominator conversation factor value: 0,02 KG = 1 EA.
But the SAP inform that is not accepted decimal.

The minimum value is equal 1?

tks

[varunanand]

Hi Jay,

You shold be able to achieve this functionality by leveraging data blending. 

Blend data in a visualization: SAP Lumira 1.28 - YouTube

Thank you,

Varun Anand