Desirability

[#1]

Dear all,
I performed a DSD for my 8 factors and 3 responses and started analyzing my data using the fit definitive screening design platform. For each of my responses, I have a model and a desirability function for each model.
Is there any possibility of maximizing the total desirability for 3 responses simultaneously,
many thanks in advance

[#2]

Hello @ELH,

Yes, you can optimize the total desirability for several responses simultaneously, by clicking on the red triangle next to Prediction Profiler and clicking on "Optimization and Desirability" -> "Maximize Desirability".

By default all your responses will have an importance of 1, so JMP tries to equally optimize the responses, but you can change the importance, goal and ranges for each response by clicking on "Optimization and Desirability" and then "Set Desirability".

More infos in the documentation : Desirability Profiling and Optimization (jmp.com)

For DSDs, you can save your prediction formulas (Prediction Formula + StdErr Pred Formula to have the confidence range)  in the datatable, and then re-create the profiler in "Graph" -> "Profiler", specifying your prediction formulas for your 3 responses (Pred + StdErr Pred), and then do the optimization from this recreated profiler. You'll have a complete view on your three responses and they will be considered at the same time for the optimization (compared to single profilers done in the traditional "Fit DSD" platform).

Hope this will help you,

[#3 · Accepted solution]

Sorry for the double post/reply.
I just checked on the Fit DSD platform, there is an easier way to have access to the full profiler.

1. When you're on the Fit DSD platform with the different effects detected, you can press CTRL+"Run Model" (see Fit_DSD screenshot) and this will open the Fit Group platform with the different effects detected for your responses.
2. Then, you can click on the red triangle next to "Fit Group" and click on "Profiler" and the complete profiler will be accessible to test your optimization of the three responses !

I think this method is more straightforward and easier to do when using the Fit DSD platform.
Hope this will help you,

[#4]

Dear @Victor_G ,

Thank you for your feedback, your suggestion works very well.