2 Replies Latest reply on Aug 2, 2018 5:50 AM by harald.solhaug

    Using a box plot with discrete data

    harald.solhaug
    harald.solhaug Aug 1, 2018 10:52 AM

      Hello!

       

      I'm having a bit of trouble with a box plot here.

      I believe the problem is that I'm using "discrete" data on the Y axis. The box plot then ends up with a "snap to grid" look.

      I don't think this is correct, unless I have misunderstood how box plots work. The average value for instance, is not expected to be an exact whole number from a random group of measurements.

      I've included a picture of how my plot looks, as well as some random data in excel.

       

      Thanks for any help!

       

      Capture1.PNG

        • 1. Re: Using a box plot with discrete data
          lei.chen.0
          lei.chen.0 Ambassador Aug 1, 2018 9:29 PM (in response to harald.solhaug)

          Hello harald,

           

          I don't think there is anything wrong with the appearance of your boxplot.

           

          One thing I'd like to point out is granularity.

           

          I added one column [#] into your data, which is simply a unique index.

           

          [#] is placed onto "Detail" card, thus each point in the plot stands for one row in your data.

          Since some of the rows have the same "Y axis" value, some points are overlapping each other.

          image.png

           

           

          Regards

          Lei

          • 2. Re: Using a box plot with discrete data
            harald.solhaug
            harald.solhaug Aug 2, 2018 5:50 AM (in response to lei.chen.0)

            Hi,

            Thank you, you are of course correct!

             

            My knowledge of boxplots was simply a bit limited. Turns out they are not based of averages, but medians. So when you use discrete data, they will always get this appearance (and box plots probably aren't very useful in a case like this).

             

             

            Best regards,