Subscribe Bookmark RSS Feed

JMP DoE Mixtures - input previously known results

anderboski

New Contributor

Joined:

Jul 6, 2017

Hello everybody,

 

I am new to jmp and I have a question for you all. I am trying to do a doe to know which components of a mixture have the biggest effect on one of its physical characteristics. This components appear in different % (in between a range) for different finished products. I have designed a model using this property as the Y and assigning the predict goal (as I can measure it).

As I already have some measurements for different concentrations of these components, I was wandering if there is any possible way to input this data into the model for it to have more information and maybe to reduce the number of needed additional measurements.

 

Thanks for your help!

1 ACCEPTED SOLUTION

Accepted Solutions
markbailey

Staff

Joined:

Jun 23, 2011

Solution

There are a couple of ways to accomplish your goal but I think that the easiest way is to create a custom design for these mixture components and then replace the runs in the resulting design with the combinations you already have. You can then augment this design with new runs that will complement the existing data.

JMP creates meta-data at the column and table level when you design an experiment and make a table for it. The other way is to enter the existing data and create the meta-data by hand. I have done this many times but it is tedious work so I suggest the first way.

I think it is a great idea to re-use existing data as long as it can be qualified for the new purpose. The existing data might be inadequate for fitting the desired model but augmenting will improve the data considerably.

Learn it once, use it forever!
3 REPLIES
markbailey

Staff

Joined:

Jun 23, 2011

Solution

There are a couple of ways to accomplish your goal but I think that the easiest way is to create a custom design for these mixture components and then replace the runs in the resulting design with the combinations you already have. You can then augment this design with new runs that will complement the existing data.

JMP creates meta-data at the column and table level when you design an experiment and make a table for it. The other way is to enter the existing data and create the meta-data by hand. I have done this many times but it is tedious work so I suggest the first way.

I think it is a great idea to re-use existing data as long as it can be qualified for the new purpose. The existing data might be inadequate for fitting the desired model but augmenting will improve the data considerably.

Learn it once, use it forever!
anderboski

New Contributor

Joined:

Jul 6, 2017

Ok that works! thanks!
markbailey

Staff

Joined:

Jun 23, 2011

See Help > Books > Design of Experiments. This free JMP guide contains a lot of information about DOE. I recommend that you read the first chapter for an overview of custom design. You will also find a chapter devoted to mixture experiments.

We also offer complete training for the design of experiments. You can review the JMP training curriculum here.

Keep posting questions - we're here to help.

Learn it once, use it forever!