0 Replies Latest reply on Nov 2, 2018 4:05 AM by alice scarduelli

    Need to calculate Regression Coefficients in a Non-Aggregate way

    alice scarduelli

      Dear Tableau Community,

       

      In my analysis I need to calculate the coefficients from the regression of Y over X (Y=alpha + beta*X) , and need to use these coefficients in a further calculations to make predictions on new data.

      This means that I will have to multiply the calculated coefficients with a new variable X_new (Y_new=alpha + beta*X_new).

       

      I successfully calculated the coefficients as            

       

      beta= WINDOW_SUM(AVG([X])-WINDOW_AVG(AVG([X])))^2   /   WINDOW_SUM((AVG([Y])-WINDOW_AVG(AVG([Y]))*(AVG([X])-WINDOW_AVG(AVG([X]))) =

      alpha= Y-X*beta

       

      This formula calculates correctly the Slope and the Intercept but it does so in an Aggregate way, so that if I now want to calculate Slope*X_new the calculations gives me the error

      "cannot combine aggregate and non-aggregate measures in the same calculations".

       

      As aggregating X doesn't make sense for my analysis I was wondering if there's a way to disaggregate Slope and Intercept, or if there is another way to calculate them directly as a non-aggregate field. I have been trying several methods and haven't succeeded so far.

      Any help would be much appreciated!

       

      Thank you,

      Alice