319607_en-US

liliconcom

Custom design from candidate points

There is two main schools of how people can build DOE. The one chosen by JMP is to create design matrix using the powerful coordinate exchange algorithm. The second way is to provide a list of candidate data points (that could respects some complex constraints) and then search through it to create the design. (Actually, is JMP creating this list of candidate points in background ? When looking at design, there are some weird non rounded values, some times, which seems to indicate that JMP can take really any point (satisfying constraints if any) from the space of factor.)

 

I discussed yesterday with many DoE users at a conference, and they told me the fact that the second option does not exist in JMP is one of the main reason they don’t want to buy and use JMP. This option is however available in SAS proc Optex, which I am sure share many similarities with the Custom DoE platform of JMP.

 

Would you think such addition may be possible in the future, for Custom Design and Augment Design platforms ?

Design of ExperimentsRe: Custom design from candidate points - Status changed to: ArchivedWe are archiving this request. If this is still important please comment with additional details and we will reopen. Thank you!Re: Custom design from candidate points

Hello,

 

Thanks Phil, Excellent news. This could be made more visible. I can't wait for V16 then

 

Enjoy,

 

Pierre

Re: Custom design from candidate points

Yes, you can design an experiment with factors only from the candidate table, if you wish. 

 

Look out for changes to the interface for this in the next version of JMP.

Re: Custom design from candidate points

Hello,

 

Thanks for this. It seems it can allow to use only covariates then (use the table as candidate list) and I can do that without adding any factors from the ones available in the candidate list.

 

So it helps, even if it was not clear from the help

 

Pierre

Re: Custom design from candidate points

Have you seen that you can define Covariate factors in Custom Design? https://www.jmp.com/support/help/en/15.2/index.shtml#page/jmp/experiments-with-covariates.shtml

This enables you to specify factors for the experiment from a table of candidate runs.