3 Replies Latest reply on Dec 30, 2016 10:14 AM by diego.medrano

    Confusing aggregation when working with a cube data source

    Benjamin Chen

      Hi, all

       

      I have encountered a problem that I don't know from where it resulted, Tableau or my cube data source(SAP BW).

       

      To make my explanation simpler, I put it in a more basic way.

       

      Say, I have a dimension [ZJHND] which contains years from 2014 to 2017.  Based on this dimension, I created a few calculated members which simply pulls each of the year level from the original dimension.

       

      5b338b8d.bmp.bmp

       

      When I dragged this [ZJHND] and some measures into a view. I found out that one measure has different values for the primitive members and members I'd created!

       

      For example, the first measures takes a value of 65.00 for 2014 and 33.00 for New2014. Some measure may behave more normal, such as the 4th one, taking same value for xxxx and Newxxxx. We did this calculation in the data source(SAP BW), it shows that results for the Newxxxx are correct.

       

      8271a4f6.bmp.bmp

       

      What is more amazing is that when I filter on this particular dimension or drag in another dimension, those values change! Values for members xxxx became the same as those for members Newxxxx!

       

      When I select 2014 and New2014 in the filter, the first measure takes same value for both members!

       

      WHY!!!! Where does the problem come from? From the way Tableau manipulates cube data? Or from the way my cube aggregates the raw data?

       

      a5de4200.bmp.bmp

       

      If anyone has encountered a similar issue, please provide me some guidance to figure this out.

       

      Thanks a lot!

       

      Benjamin