7 Replies Latest reply on Feb 15, 2013 8:22 AM by Jonathan Drummey

    No exclude/keep?

    Shawn Wallwork

      When you turn off "Aggregate Measures" both Exclude & Keep disappear in the tool-tips and the rt-click context menu. What's the thinking behind this? There are times in a scatter plot I want to exclude an outliers :

       

      Exclude.png

       

      But since this chart is not aggregated I can't. Wondering why? Thanks,

       

      --Shawn

        • 1. Re: No exclude/keep?
          Shawn Wallwork

          Bump & call out: James, Dan , Robert.

          • 2. Re: No exclude/keep?
            Robert Morton

            Hi Shawn,

             

            I believe the reason has to do with how duplicate values are handled. Consider the following data set.

             

            x,y

            1,1

            1,2

            2,1

            2,1

            2,1

            2,1

             

            In a disaggregate scatterplot, the tuple (2,1) will appear as multiple marks stacked on top of each other. You can select the top-most mark, or you can lasso to select all four. Each mark is separate, but all four would be eliminated if you tried to exclude the top-most mark, since all four would be filtered by excluding x,y=2,1. Our include / exclude experience is driven by a specific mark or collection of marks you have highlighted, and we do not wish to eliminate marks which have not been selected. Since we cannot guarantee that behavior in disaggregate mode, we don't offer the menu item.

             

            Does this help?

            Robert

            1 of 1 people found this helpful
            • 3. Re: No exclude/keep?
              Shawn Wallwork

              Robert thanks for the explanation. I appreciate understanding your thinking.

               

              --Shawn

              • 4. Re: No exclude/keep?
                Robert Morton

                Glad to help!

                -Robert

                1 of 1 people found this helpful
                • 5. Re: No exclude/keep?
                  Jonathan Drummey

                  Thanks for the explanation, Robert! One of the things I really appreciate about Tableau is the way you (developers) give thoughtful responses to our questions.

                   

                  I played with this one a bit because I was wondering if there was any difference between using disaggregated marks (which I hardly ever use) and dimensions (which is what I more commonly use when I want to plot every mark/value). What I found was that the behavior that Shawn observed and Robert described is the same for both, which makes sense given the explanation. The one difference (that I expected to see) is that a table calc cannot use a disaggregated measure for partitioning or compute using, but does change results because all the rows are being returned, whereas a table calc is of course affected when using dimensions. One potential bit of confusion in the UI here is that pill looks the same for a continuous dimension and a disaggregated measure, the only way to tell the difference is by looking at the pill's context menu.

                   

                  Shawn, my suggestion to get the Keep Only/Exclude would be to get enough dimensions in the level of detail of the view so you can use ATTR() as an aggregation. I prefer ATTR() in this case to MIN() or MAX() because it returns Null if there are multiple values of the given measure, which would be the case if the level of detail was at too coarse a granularity.

                   

                  Jonathan

                  • 6. Re: No exclude/keep?
                    Richard Leeke

                    This one caught my eye too, and I also played with it because I never think to use disaggregated measures - like you I always tend to convert to continuous dimensions.

                     

                    My observation was slightly different to yours, Jonathan. You said:

                     

                    " I was wondering if there was any difference between using disaggregated marks ... and dimensions... What I found was that the behavior that Shawn observed and Robert described is the same for both".

                     

                    Here's what I see when I turn off aggregate measures, just like Shawn's original question. Note the lack of Keep Only/Exclude and the count of 4248 marks:

                    Disaggregated.PNG

                    And here's what I get with continuous dimensions. Keep Only/Exclude has come back and I only have 3304 marks:

                     

                    Dimensions.PNG

                    This all fits with Robert's explanation, of course. In the latter case there is no ambiguity about which mark you want to keep or exclude, since there is only one mark for each distinct permutation of dimension values.

                    • 7. Re: No exclude/keep?
                      Jonathan Drummey

                      Hi Richard,

                       

                      I think I know the difference here: I was basing what I was doing on the the simple 6 rows of data Robert had posted and switching from an aggregated measure to a continuous dimension changed the view to the same level of detail as disaggregating the measures, so it was showing 6 marks. I added a dimension to the data at a coarser level of detail and can duplicate the behavior you describe, see the attached for details. This is a great example of how it's important to know the data!

                       

                      Jonathan