11 Replies Latest reply on Apr 5, 2013 12:58 PM by Nathan Lee

    Modified Z-Score? (Requires Median Average Deviation)

    Nathan Lee

      Greetings. I am trying to implement a Modified Z-Score (http://www.itl.nist.gov/div898/handbook/eda/section3/eda35h.htm#Z-Scores) in my data analysis. The data I am looking at may be significantly affected by extreme outliers, and the Modified Z-Score is better suited than the "standard" Z-Score. I have already managed to implement the Z-Score calculation.

       

      The issue I am running into is the Modified Z-Score's use of the Median Absolute Deviation (MAD). To calculate it, I need to calculate the median of the difference of each data point from the median of the data:

       

      But I can't seem to calculate the median of differences, because Tableau tells me I can't calculate an aggregate function on an aggregate.

       

      Has anyone here implemented these functions?