3 Replies Latest reply on Oct 25, 2016 12:21 PM by David Li

    Data blending issie

    Amy Amy



      I have 2 data sources, table one (A) has employee detail (name, id, region, department etc.) and table (B) has the type of mobile device some employees use (date purchase, type of OS etc). Only 50% of the employee has laptops listed in table B.

      i am blending two table using employee id to get the count of laptop by OS type which then can be filtered by region. Everything is fine up to this point. But i cannot get it right when i want to get the usage, for example number of mac users over all employees in table A.  50% of users in table B are mac users so the answer should be 25% but i am getting 100%. When i remove the data relationship then i get the correct  (25%) but then I cannot filter by region which is in table A. I have used FIXED syntax but no luck, the calculation is correct but when  i move it to the mark, it become red (error).


      Total employees=1000

      Mac users = 250

      % of Mac users out of total employees = 25%

      Count of employee --> {fixed [employee ID]: sum([employee count]}

      Any help with this is greatly appreciated.



        • 1. Re: Data blending issie
          David Li

          Hi Amy, you won't really be able to use a LOD calculation here because a data blend can only give you aggregated values, and the secondary data source can't contribute to the level of detail of the sheet.


          What version of Tableau are you using, and what kind of data source? If you can use Tableau 10, this will be pretty easy to solve with cross-database joins.


          Also, do you have a packaged workbook you can share?

          • 2. Re: Data blending issie
            Amy Amy

            Hi David,


            Thank you so much for the respond. I just updated to version 10. By using cross-database joins and fixed syntax i was able to get the expected outcome.


            Thanks again!


            • 3. Re: Data blending issie
              David Li

              You're very welcome! If you wouldn't mind, could you please mark the reply as correct so this no longer shows up as unanswered? Thanks!