1 Reply Latest reply on Aug 16, 2016 8:26 AM by Frank Casella

    Filter vs Parameter for compounding data

    Frank Casella



      I'm struggling to make a workable dashboard with cascading selectors.  The dashboard shows a geographic view with the following user choices:

      • 13 different metrics split into three groups (current, 6-month lagging, and 12-month lagging indicators)
      • 3 different coloring options (result, point change from prior year, percent change from prior year)
      • 49 different demographic "filters" split into ten groups


      I'm using parameters for the "filters" because some of the demographic views overlap.  For example, there's a filter for home ownership of <1 year, <3 years, and <5 years where <3 years is calculated using the base data characteristics of <1 year and 2-3 years

      Case [Parameter]

      WHEN '<1 year' THEN

      (IF [SEGMENT] = '<1 year' THEN [NUMERATOR])

      WHEN '<3 year' THEN

      (IF [SEGMENT] = '<1 year' OR [SEGMENT] = '2 to 3 years' THEN [NUMERATOR])



      The problem I'm running into is when I try to create the dashboard.  I can't set the parameters to show relevant values only, so I've tried to use cascading parameters:

      Case [Parameter Group]

      WHEN 'Housing' THEN

      CASE [Housing Parameter]...

      WHEN 'Age' THEN

      CASE [Age Parameter]...



      Everything works from a calculation perspective, but if I show all of the parameters on the dashboard (14 plus a couple context filters), the dashboard is very cluttered.

      • I can't use "pops" to hide and show the parameters, because the user will only be able to select whichever parameter is on top and can't click through the container.
      • I can't switch to filters because of the compounding of the base data.
      • I can't use URLs because we don't yet have server.


      Is there "another way"?


      Thanks for your help.



        • 1. Re: Filter vs Parameter for compounding data
          Frank Casella

          Hi everyone,


          Never got a response, but I figured a way around this.  It required creating a couple additional data tables outside of Tableau to make it work.  Here's what I did (in case anyone else ran into this problem or was curious.

          • Duplicated the original data tables to create the needed aggregate denominators outside of Tableau.  This allowed me to create "group" denominators for the demographic segments that would ignore any demographic filters.
          • Created a relationship table to left join all other tables.  This prevented data permutations and ensured aggregate lines were calculating correctly.
          • Used a parameter to filter the "group" and a quick filter to filter the demographic segment
            • created a "segment filter" calculation [Parameter].[Segment Picker]=[Segment Group] for the "group" filter
            • set the quick filter to show only relevant values

          Since the totals table didn't have a demographic segment field, it was unaffected by any demographic segment filters.