cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
JMP is taking Discovery online, April 16 and 18. Register today and join us for interactive sessions featuring popular presentation topics, networking, and discussions with the experts.
Choose Language Hide Translation Bar
Shannon
Level III

Custom DOE, 4 continuous factors, want to add replicate runs

I am creating a DOE with 4 continuous factors, including 2nd order interactions and quadratics. I would like to add replicate runs. 

 

I tried typing 5 into the replicate runs box before creating the design, but the design that was created did not have any replicates. The number of runs was 20 and all were unique. I understand replicates to be running the same set of conditions more than once.

 

Am I misunderstanding what a replicate run is? or am I setting up the DOE incorrectly when I create it?  

 

 

 

1 ACCEPTED SOLUTION

Accepted Solutions

Re: Custom DOE, 4 continuous factors, want to add replicate runs

I'm very sorry you ran across this issue, and sorry I did not see this thread until now.
Any chance you're using 14.0? This was an issue that should be fixed in JMP 14.2 (and 14.1). In 14.0, you will have to add the replicates via jsl.

Cheers,
Ryan

View solution in original post

20 REPLIES 20
Peter_Bartell
Level VIII

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Could you attach the JMP data table containing the design? I just tried to recreate your design and JMP did indeed give me 5 replicate rows in the data table...they all aren't the same treatment combination because in the custom design platform the coordinate exchange algorithm is maximizing your optimality criteria by placing the replicate points in the 4 factor coordinate space to maximize the optimality criteria, not just blindly give you identical treatment combinations for your 5 replicates. If you want to 'force' the data table to have 5 identical rows...I suggest making a 15 run design and then add your replicates (identical treatment combinations) by hand...but that won't maximize your optimality criteria and will also influence other design evaluation critieria as well.

Shannon
Level III

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Here is an example data table with factor and response names removed. I have included 5 replicates out of 20 total runs.
Can you explain a little more which of these you would consider replicate, or what you mean by placing the replicates in 4 factor coordination space? What makes one run a replicate of another?
Shannon
Level III

Re: Custom DOE, 4 continuous factors, want to add replicate runs

 
Peter_Bartell
Level VIII

Re: Custom DOE, 4 continuous factors, want to add replicate runs

All optimal DOE tactics/algorithms (which is the basis of JMP's Custom Design platform) use a technique called coordinate exchange to identify positions in the factor design space to achieve a goal of the specific technique. Generally speaking it's to maximize what is called the 'optimality criteria'. There are different types of optimality criteria...and JMP supports most...such as D, A, I etc. Each has some really elegant math behind them. Explaining how this works in a discussion forum such as this is beyond the scope of the forum. I suggest taking a look at a great book co-authored by my colleage Brad Jones and Chris Natchsheim called "Optimal Design of Experiments: A Case Study Approach" for more examples and details.

Peter_Bartell
Level VIII

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Shannon: See the attached JMP data table for a 20 run, I-optimal design, RSM model, with 5 replicates. I've assigned row markers for the replicate treatment combinations so you can find them in the randomized order. Take a look at the embedded DOE Dialog script in the data table to see the script selections from the Custom Design window. It looks like you weren't making the appropriate selections in the Custom Design planning window to generate your design.

Shannon
Level III

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Peter, Thank you so much for providing the data table.

I opened your table and each time I run your dialog I get replicates, but whenever I start a new DOE dialog it is not working for some reason. Is there some preference that I may have set differently that is causing replicates to not happen? I enter the 5 replicates in the Number or Replicate Runs like you have on your table, then after I make the data table and open the dialog again (like I did with your file) the replicate number is back to 0. The only difference I can see is that the minimum runs I get is 15 whereas yours has 20. Thanks again for your help. 

Peter_Bartell
Level VIII

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Here is what my DOE dialog window looks like...from here I just hit Make Design and the design is created...does your's look like this from a new dialog window?

Shannon
Level III

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Here is what mine looks like (capture). Everything is the same except for the order of the model terms and the minimum runs required. Then I click Make Design and there are no replicates in the design (capture 2). 

Peter_Bartell
Level VIII

Re: Custom DOE, 4 continuous factors, want to add replicate runs

Hmmm. I'm not sure what's going on. One thing you might want to check are your JMP preferences in the DOE preference group. I've attached what mine look like. Maybe the optimality criteria option is forcing a specific optimality criteria in your's? Where mine is picking the most maximal as 'recommended'?