Subscribe Bookmark RSS Feed

Factor is categorical and continious at the same time

desperado7

Community Trekker

Joined:

Sep 7, 2011

Hello,

I want to design an experiment with only two factors.

But both factors are as well categorical and continious.

Factor A, for example, is between 0 and 40 categorical (only stepwise changeable from 0 to 40) and between 40 and 120 continious

Factor B, for example, is between 0...60 categorical and between 60 and 90 continious.

How can I realize that in a DOE.

Thank you for your answers! 


3 REPLIES
paigemiller

Community Trekker

Joined:

Apr 12, 2012

This is a very vague question, and no one can simply take this information and create a DOE.

You need to state the goals and requirements of the DOE (things like what model are you trying to fit, how many total runs in the DOE, what precision do you want to have on your estimates)

desperado7

Community Trekker

Joined:

Sep 7, 2011

Firstly, Thank you for your answer.

I will try to be more precicely...

My target is that I want to keep my experiment scope as small as possible and get safe information how my factors influences my responses.

If I don`t use JMP I would make 16 experiments. Because I want to check from both factors the steps 0-40 and 0-60, and also when they become continious the borders and the points within.I also want to check hot the two factors influnce each other on the result.

0               0  60 75 90

40              x   x  x   x

70              x   x  x   x

120            x   x  x   x

Can I design a DOE with less experiments to get safe results and the most important question is how should I define my factors when they are at the same time categorical and continious.


Can I define them as continious and make the borders form 0....120 for example (for Factor A).


Sorry for my bad english...I hope it is more clear now

paigemiller

Community Trekker

Joined:

Apr 12, 2012

Actually, this doesn't answer my question. In order to design a DOE, you must specify a model that you want to estimate. Without that model, there's no answer to the question about how to create a DOE in this situation.

Regarding the continuous versus discontinuous nature of your variables, I have never seen or heard of such a thing. I suppose you have a good example of how this could be, could you share it with us?

Since DOEs all select specific levels of each variable, I don't see why continuous vs discontinuous should be a problem. You have to pick a finite number of specific levels of each factor to create the DOE.