topic Custom Design Number of Replicates in Discussions

Custom Design Number of Replicates

Zappy — Thu, 08 Jun 2023 21:08:04 GMT

Dear JMP Staff & Users,

In "Classical Design" design where replicates input (let's say a value of 3) in JMP means replicating each original runs 3 times.

For example, let's say I have 10 original runs (without replicates).

Then I input 3 replicates, the total number of runs would be : (10 original + (10original x 3)) = 40 runs.

When I use "Custom Design" , it seems the meaning of replicates is not the same as in "Classical Design".

It doesn't replicate each of the original runs.

Anyone can advise how does this "input of replicates" in Custom Design works?

Thanks.

B.r,

Chris

Re: Custom Design Number of Replicates

Victor_G — Mon, 14 Feb 2022 08:50:26 GMT

Hello @Zappy,

There is a slight difference in the terms used by JMP that explain the different results:
- In a "Classical design", JMP displays "Number of Replicates", meaning here the number times to perform each run (in addition to the original runs),
- In a "Custom design" JMP displays "Number of Replicate Runs", meaning here the number of run(s) to repeat.

Hope this will help you ! :)

Re: Custom Design Number of Replicates

Phil_Kay — Mon, 14 Feb 2022 09:25:33 GMT

Hi,

As @Victor_G has said, the definition of replicates in Custom Design is different from your definition. It is documented in the JMP Help:

"This does not replicate the entire design, but chooses the optimal design points to replicate."

Of course, you can easily replicate the entire design by copying and pasting the rows.

Regards,

Phil

Re: Custom Design Number of Replicates

P_Bartell — Mon, 14 Feb 2022 13:03:38 GMT

To add just a bit more detail to the correct explanations of @Victor_G and @Phil_Kay there are completely different DOE planning philosophies at work here between classic and optimal DOE approaches. The classic approach uses as a basis design matrix orthogonality of effects and balance as the primary principles forming the design matrix. Hence you end up with a somewhat predetermined # of runs based essentially on a desired design resolution. Replication from there almost forces the approach built in JMP's classic design platform because to maintain balance the entire design matrix needs to be replicated. Custom design (more generically known as optimal design) uses a model based approach to identifying the design matrix that maximizes a chosen optimality criteria (D, I, A, etc.). So now the DOE planning team has the freedom to pick replicate points in the space at will...based on other criteria such as material/process resources, time, etc. while still maintaining the maximal optimality criteria. Hope this helps a bit?