4 Replies Latest reply on May 8, 2018 10:39 PM by Dave Scott

    Actions and Secondary Data Sources / Asterisks in Secondary Data Source

    Dave Scott

      I'm new to tableau, and this is my first real attempt to use it for something, as the limitations with Excel are just getting to be a bit too time consuming to work around. I've built a working version in Excel using VBA and SQL, but the reporting side doesn't look great and is a pain to use.

       

      Tableau fixes so many of those issues, but things I was able to do in Excel with VBA/SQL I now just can't do in Tableau. I keep hitting limitations and hurdles with Tableau, finding a work around, just to smack into a new limitation or hurdle. Most of it has to do with multiple data sources and filtering.

       

      I'm trying to analyze a reasonably large data set, I'm then planning on distributing the readonly in reader to a few people  who can drill through the data set. The idea is to take a global set of groups, then based on some flags or general quantity of values within those groups, narrow them down to a handful of groups, selecting each one that interested in and get a more detailed report for that single group.

       

      The groups are made up of individual ID numbers, these ID numbers need to be put into a single group. This grouping however is imprecise and a bit subjective. So I want the user to be able to regroup the IDs if required. Maybe even keep their preference of groupings on file somewhere.

       

      Site number is Unique and will only be in 1 group.

      AutoGroup

      ManualGroup

      Site Number

      GroupA

       

      1

      GroupA

       

      2

      GroupB

      GroupA

      3

      GroupB

       

      4

      GroupB

      GroupF

      5

      GroupA

       

      6

      GroupC

       

      8

      GroupC

       

      9

       

      As reader can't use Live data sources, and there's no way of letting the user change the data in the reader. I've written a little excel tool to swap the data out in the exported tabeau file. This only works on non-joined data, soon as join it it's converted to tableau file format and won't work.

       

      So I've got 1 table, as above that’s kept as a single Excel Table, in the document. All it contains is ID to lookup and Group that was automatically assigned, and the users preference for what group they want it to be.

       

      My issue is data blending just seems to be half baked, and soon as start using it will break something in some weird and wonderful way.

       

      It seems to me that my main issues are, with either one of these working would solve my problems and allow for a work around:

      • Action Filters don't apply to a secondary source (but normal filters can, if temperamentally). This makes no sense to me and just seems like a massive oversight and is the root of most my problems. Is there some kind of work around that allows me to have my data tables as a primary source, but allow me to then do all my filtering and actions on this secondary source?
      • I can't use a calculated field to get a value from a secondary source and have it as a dimension. Always comes through as a measure no matter what try and then end up with asterisks all over the place if use it for anything.  Is there a way of pulling through a value into another source so that it becomes a value in the primary source, like a calculated field join or a lookup equivalent?
      • If I've got the site number (The linking ID) in use on the sheet I get a lot less issues. But that's not the desired result, as then all other items per group but over multiple sites can't be aggregated. Is there a way to trick Tableau into thinking there's a value in the table but ignore it?
      • Alternatively, I can set the Group table as primary source and then apply all filtering off that. That does then break all other action filters if I wanted to do any on the dashboards of the main data. But more of an issue, if I bring in any value as a Row I end up with asterisks everywhere again. (Again, this makes no sense - just list them all, there's a link). Is there a way of getting the Data Blend to act a bit more like a join and treat the data as actual lines of data rather then aggregating it and then giving up with the values if there's multiple?

       

      Anyone got any idea? I've been going through all the topics I can find on this, but most I've come across are unanswered with no solution. I've tried so many things, but I'm ready to throw the towel in and just head back building the solutions manually. It's more work but atleast I know there's a solution. Data blending just seems half baked and doesn't really work as intended in any way that I've tried.

       

      Thank you for any help can give.

      Cheers