4 Replies Latest reply on Apr 20, 2016 3:40 AM by Andrew Flynn

    Tableau data sources - conformed dimensions

    Andrew Flynn

      I am using Tableau to connect to a data warehouse made up of several fact tables as star schema, mostly interconnected, and mostly using conformed dimensions.

      Does anyone know what is best practice for setting up the data sources in Tableau Desktop?  I do need to connect to the whole set of tables.

      I have been moderately successful using a mixture of joins and blends, but the approach has some problems.

      These are, particularly: having multiple copies of the same dimension but under different names, when joined to different fact tables in the same data source schema; e.g. Region connected to one fact table, and a duplicate called Region1 connected to another fact table.

      The second issue is to do with blends:finding appropriate linking dimensions, and not being able to use the COUNTD aggregation for a secondary source measure.

       

      A a few specific questions about best practice:

       

      1. Is it possible to join a single dimension table to two fact tables in the same data source schema?  I mean, without having to join it twice under a different name.

      2. Would it be a clean approach to create several data sources (around 12)  each with a single fact table joined to several dimensions; and then to blend combinations of them in the worksheets?  How would I overcome the COUNTD aggregation issues in the blends?

       

      Thanks a lot

       

      Andrew