Desirability Summary from FPC Profiler

[#1]

I am exploring the Functional Data Explorer features in JMP Pro and fitting multiple models to my data where I provide a target function to feed the desirability function.

I have 10 input functions that have different desirability values relative to the target function and I would like to get a simple summary table of desiriability for each of the 10 input functions.  Is there a menu command to do this?  My manual workaround is selecting points on the Score Plot to filter the data and copy/paste the Desirability value from the FPC Profiler.

[#2]

Hi @TXJo,

Welcome in the Community !

I see one easy option that may help you collect the values without having to copy-paste manually the desirability values: click on the red triangle of FPC Profiler, go to Optimization and Desirability and select Save Desirability Formula :

This option will create a new column in your original table with the desirability formula values calculated for each of your data points:

To enable this option, you need to make sure that you have already set desirability functions, available in the same menu Optimization and Desirability > Desirability Functions and Set Desirabilities... In this JMP example dataset Formulation For Homogeneity DOE, I have set the different desirability functions for the three responses (original response, Difference from Target and Integrated Error from Target) like this to allow the calculation of Desirability formula :

If you want one desirability value for each different curve (and not for each data point), you can click in the red triangle next to Function Summaries, and select Save Summaries. You can copy the Desirability formula column obtained from the previous step and paste it in the Model Summaries table. You'll get a message error as you won't have the same column names and info in this Summaries table, you can ignore the error. Then, in the pasted Desirability formula column (right click on the column name, then Column Info), select Formula > Edit Formula, and replace in the equation the term not present in the datatable by Response Prediction Formula available in the table :

You'll then get the desirability value for each of your curves in your table:

Hope this solution will work for you,

[#3]

Thank you Victor for the clear and quick recommendation. This method is very nice but I am having trouble reproducing Desirability values based on the Integrated Error from Target (which I did not describe or show in my original post).

Full view of profiler, with lot 2887 selected:

 

Desirability coefficient table:

 

The Save Desirability Formula command writes the formula for the mean response (Size/nm) by default but my interest is Desirability of the entire function so I manually edited the formula to reflect the Integrated Error from Target coefficients.  I am not finding a column in the Summary that produces results that match the profiler.  For example the formula below results in 0.837, not 0.596 as shown in the profiler graph above.

Is there a way to manipulate the Summary table data to reproduce the Profiler's Desirability value for Integrated Error from Target?

[#4]

The desirability values you're trying to match are the ones which are time dependant (the first variable of the FPC Profiler). The first method I described should give you the right raw desirability values for each data points (with time dependancy), whereas the second method will give you a desirability "summary" for each curve.

As time has no influence on the integrated error from Target, the first method should give you the same values as in the Profiler.

Hope you'll be able to match the values and solve your problem,