Ternary plot (with more than three components)

[#1]

Hi, how can I visualize aTernary plot (with more than three components) using color gradients  to show the variation of the response

[#2 · Accepted solution]

Hi @hamza-moussa,

Welcome in the Community !

To visualize the variation of the (predicted) response in regards to the mixture factors, you have to model the response and save the Prediction Formula (from the "Fit Model" window, click on the red triangle, "Save Columns", "Prediction Formula") :

Then, in the menu "Graph", use the "Ternary plot" option, and specify your mixture factors as "X, plotting" and your prediction formula for your response as "Contour Formula" :

You'll get to the ternary plot, and from there, you can click on the red triangle and customize the graph :

In this example, I changed the color theme to a chromatic red-orange/yellow-green them, and choose the option "Contour Fill", "Fill Above". I now have this graph that should be similar to what you are expecting :

You can also change the different levels/values for the prediction formula if you want to change the space between each individual levels.

The example is "Elongation of Yarn", from https://online.stat.psu.edu/stat503/lesson/11/11.3/11.3.3

If you have more than 3 factors, no problem, JMP will create a grid showing each time 2 mixture factors and the last factor will be the sum of the remaining mixture factors :

 

I attached the datatable with the model script, prediction formula, and the ternary plot script so that you can have a look at it.

I hope this example will help you,

[#3]

Thank you so much Mr. @Victor_G 

[#4]

can we do the same thing for Mixture profiler ?

[#5]

Yes the Mixture Profiler is the perfect solution to visualizing ternary plots interactively when there are more than 3 variables.

[#6]

Thank you @louv, can we make mixture profiler using gradient color ?