Backtransformation of Y on prediction profiler

[#1]

Hi, 

After having transformed the Y data to fit a model of a DOE, how can I generate a prediction profiler that has the actual predicted Y on the Y axis instead of the transformed Y? I would like to backtransform it for ease of interpretation. 

Thank you,

Sara

[#2]

Hi @SaraA ,

You can save the prediction of the transformed Y in the Fit Model and then create a new column in your data table and add a formula back transforming the saved prediction formula to the original Y. This column can then be used in the profiler (Graph>Profiler) to visualise.

Thanks,

Ben

[#3]

Hi Ben,

I have tried this but when I try to generate a profiler with the backtransformed data, it does not contain the factors of my original model. It just plots the Predicted Y on the X axis and the backtransformed predicted Y on the Y axis.

[#4 · Accepted solution]

Hi @SaraA ,

Apologies, if you bring in both of the columns it should come up in the Profiler.

If you put both of the formula columns in the profiler option it should show the relationship of both

[#5]

Hi Ben,

The problem is that this is what I get when I ONLY select the backtransformed column:

I don't want this profiler. I want the backtransformed data on the Y axis and my factors (I have 5 of them in my original model) on the X axis. Like this:

I don't know why the profiler does not show this factors when I use the backtransformed column.

[#6]

Hi,

I found out how to fix it, thank you!

[#7]

Hi @SaraA,

If you have a backtransformed formula column of Y (based on the prediction formula column Y found with the model), selecting this backtransformed formula column of Y and checking in Profiler "Expand Intermediate Formulas" will display only the original predicted Y depending on the factors (and not depending on the transformed Y).

Hope this may help other users facing a similar problem,