3 Replies Latest reply: Jun 29, 2011 3:56 PM by John Witherspoon RSS

    Linear Regression lines inside a scatter chart

      I have a scatter chart as --

       

      Dimension -- Client_Name  (dots on the chart)

      X-Axis -- Revenue

      Y-Axis -- ACI

       

      Now I need to add 2 linear regression lines in the chart. I tried adding a new expression using linest_m function and changing the display option as line but for some reason, the regression line is not being displayed..

       

      Am I missign anything? If anyone has a sample prototype that shows scatter chart with regression lines and can share it, that would be very helpful!

       

      Thanks in advance!

        • Linear Regression lines inside a scatter chart
          John Witherspoon

          For a linear trendline, checkmark "Advanced Mode" on the Expressions tab, and then checkmark the linear trendline.  I don't know if that's what you mean, though.

           

          I thought I had an example of calculating some boundary lines in a scatter chart, but I can't seem to find it.  Maybe I'm misremembering.

            • Re: Linear Regression lines inside a scatter chart

              John, thanks for quick reply!

               

              Basically, I am trying build a chart as per attached imaeg... the image is a mockup designed in Excel.

               

              The dots indicate - Client Names (thats the dimension for this scatter chart)

               

              and the 2 expressions are - revenue and ACI

               

              So far I am able to prepare the chart with client name dots and revenue and ACI being on X and Y axes.

               

              Now as per QlikView documentation (section 21.4) about linear regression, I added 3rd expression as below --

               

              =linest_m(total aggr(column(2),CMU2_NM),CMU2_NM) * CMU2_NM

              + linest_b(total aggr(column(2),CMU2_NM),CMU2_NM)

               

              This obviouskly didnt work... my dimension represents a client name so not sure how I can apply the multiplication in above formula...

               

               

              Now in terms of Trendlines, I wont be able to use that because I need to add 2 seperate linear regression lines... one for all the clients and another one for clients selected by filters applied..

               

              Appreciate any help...