1 Reply Latest reply on Jan 19, 2014 10:07 PM by Prashant Sharma

    Brand New to Tableau :(    Z-score with moving average?

    C S

      Hello so apologies in advance as I am a quite new to Tableau (first week on the job!). This probably elementary to you all but I could use some help. Already used up almost a day thinking about this.

       

      I am not even sure if I am thinking about this correctly so feel free to help change my approach if it is better than my current one. FYI Sum Total = sales. I just wrote it as such since i'm combining diff prod. into one.

       

       

      So I have sales and date data.  I am aiming to track sales behavior to make sure it is moving along what is expected (prior year) and ultimately signal to me whether sales are unusually low or high(outliers)

       

      1. Compare year over year on a weekday scale (aka 1st Monday 14" compared to 1st Monday of 13")  - I think I can do this.

       

       

       

      2. Create z-score on moving average of dates.


      I am not too familiar with the calc fields but I made it as:

       

      Z-score:

      (AVG([Sum Total]) - WINDOW_AVG(AVG([Sum Total])))

      /

      WINDOW_STDEVP(AVG([Sum Total]))

       

      Moving avg:

      WINDOW_AVG(SUM([Sum Total]), -2, 0)

       

      not sure how I go back inserting the MA into the z-score calc since I cannot aggregate the aggr

       

      NOTE: Can somebody confirm if I even have the date orientation achieved correctly?

       

       

       

      3. Once I have the above score.settled I could use suggestions on presentation.

       

      What I mean is at the dashboard level I want the list of clients that have the unusual behavior to show up (big or small z-scores). Do I need to know anything about the above that may limit this presentation??