0 Replies Latest reply on Jun 4, 2012 10:03 AM by jeff.vitek

    Improving Performance with sorting



      I have only been working with Tableau 7.0.0 for a few weeks now, and I've encountered a sorting issue that I can't seem to find a good solution.  I am in the midst of an internship visualizing billions of rows of data on a greenplum server using Tableau.  It is all medical data, and therefore I cannot attach any example data sets, so I will try to explain my issue with as much detail as possible.


      I am attempting to find a specific insurance carrier's top 25 referring doctors and display the top 5 referred sites for each of those docs in a specific US state.  I have been using the "Display top N within a category" article to duplicate & create a set, but that takes 5 minutes, then creating a filter can take 5 additional minutes, and finally sorting twice can take another 15 minutes to execute.  I need them in descending order (number of records -- carrier & state filtered) for physicians and site.  In other words, I want to see the top referring doc and his favorite site at the top, then his 2nd-5th favorite sites, then the #2 referring doc and so on as a bar graph.


      I have filtered the insurance carrier name and state with context filters and I can get everything sorted properly using sets, but the load times often lead to "Tableau is out of memory."  Unfortunately, the data needs to be connected live so that the workbook can be useful in the future looking at daily changes.  Is there a way to implement the filters on the data and have that data "saved" so that additional filters only look at the already filtered data rows?  That would cut the query down from billions of rows to a few million rows.  I am also cleaning up the data slightly by grouping multiple site names that are referring to the same place IE: "Jeff V's Care Center" and "Jeff V's Care Center (RCS)" when they are at the same address.  This has proven to run Tableau out of memory as well.


      Thanks for the help