1 Reply Latest reply on Nov 19, 2018 5:13 AM by Yuriy Fal

    Shared Dimensions in Tableau

    Alexandru Vadenau

      Hi there. Ill shortly present my case and then ask for the questions.


      In our project, one objective we had to attain was to develop using cognos a model of shared dimensions. What this shared dimensions means is the following: we had several data marts and for each of them we created a separate package. Carrying on, the business wanted to be able to have a package with information from all the datamarts and to be able to have this power package that would let them see certain values. We managed to do that in Cognos as we basically didnt form the snowflake around the fact table, but we took all the dimensions wanted in this Shared Dimensions package and for each of them we formed snowflakes with fact tables from all the involved data marts. So basically we had each dimensions joined with 6-7 fact tables.


      Now, the project shifted towards Tableau. From what i have observed in Tableau Desktop ( as this is the only tool we have access to) , there are 2 ways of combining data. One would be data blending, but that doesnt really fit our needs as one limitation of the data blending would be that when bringing information from the secondary data source, it will just pop the results that match the common field in the primary data source, which is not particularly what we need, as what our business needs is to have almost information mapped even though there is no link between the information dragged in the report by the user. The second method i have investigated in Tableau would be cross database joins and here is where i think i would need suggestions if you think Tableau can manage our need. So, from what i have seen , we can start with a fact table, connect one dimension and then if we grab another fact table that also has a key to that dimension, it will be nested somehow on the right of the dimension. Basically, when i add a new fact table to the construction, i dont want it to be somehow put to be joined with the original fact table. In the end, this is my question. Can this be done in Tableau?


      Thanks a lot for your help in advance!



        • 1. Re: Shared Dimensions in Tableau
          Yuriy Fal

          Hi Alexandru,


          Though Tableau Datasource couldn't have

          a complex entity-relation data model (yet),

          one could be using a star / snowflake schema

          (built around each fact table) as a distinct datasource,

          then combine the views and filters from (some of) them

          on a dashboard -- using both Cross-Datasource Filters

          and/or Filter / Highlight Actions on the Shared Dimensions.


          Besides, when analysing data from different sources,

          the data blending is a common approach,

          but it is mainly for the different scenarios,

          such as Actual vs Target.