Mixture design with numeric (continuous) responses and a nominal (yes/no) response

[#1]

I have recently conducted a mixture design on a sealant formulation where I have measured the tensile strength and elongation (continuous numerical response) but also recorded if there was plasticizer migration after curing (nominal response: yes/no). For the tensile properties I used a standard least square model, the plasticizer migration was fitted with a nominal logistic model. 

I am looking for advice how to link both models. More specific, when I use the least square model to calculate a formula with a high elongation, I would like to exclude formulas that have a high likelihood of plasticizer migration.

Is there anyway to achieve this in JMP? I am working with JMP 18.0.1

Thank you!

[#2 · Accepted solution]

Hi @SamVA,

There is a possibility to do this using formula from the models you have trained :

1. Save Formula from your Least Squares (for continuous responses) and Logistic (for your nominal response) models.
2. Use the Prediction Profiler Platform (from Graph menu, "Profiler") and use the probability formulas for the different classes of your nominal response as well as the prediction formula for your continuous response(s) (you can also save PredSE formula for your continuous response(s) to have confidence intervals in your Profiler for these responses) : 
   
3. You'll then get a common Profiler for your nominal response and continuous response(s), and you can specify the desirability of each class for the nominal response to do the optimization :
   

Hope this answer will help you,

[#3]

Thank you very much. This is exactly the answer I needed.