Subscribe Bookmark RSS Feed

Newbie question. Can I use my own data for modeling instead of using matrix created by DOE in JMP?


Community Trekker


Jul 10, 2015

Hello all,

I am new to JMP, and I only read some tutorial documents about DOE. My question is on how to use my previous data into JMP DOE modeling. I am running my experiment which depends on 6 variables. I run each variable experiment separately. So far I have done tests on 4 variables. And now I got this JMP software. Can I input my data from 4 variable experiments and run prediction modeling? If I follow DOE tutorial and created DOE matrix for 4 variables, and they give different testing parameters. So I don't know how to incorporate my current data into JMP modeling.

If someone can give me a link or some suggestions, I would appreciate.



Jun 5, 2014

sushiz: In a very general sense, from the little bit of information you have provided, you can use any number of modeling methods native to JMP or JMP Pro. From the sounds of it what you've done is collected what I call 'happenstance' data, not following a more systematic DOE approach. The reality is this happens every day in industry, society, health care, government, etc. when processes unfold as either nature or human's dictate with no regard for the desirable properties of a DOE based data collection approach.

There are a wide variety of modeling methods that can be used in this context. This sort of modeling is JMP Pro's sweet spot. Modeling methods in JMP Pro such as the collection of penalized regression based methods in the Generalized Regression platform might be very helpful. Or using the Partial Least Squares platform? Another suggestion is to watch any number of the Building Better Models or Mastering JMP, JMP On Demand webcasts too at this URL:

On-Demand Webcasts | JMP

And, oh by the way...if possible I'll also recommend using a DOE based approach for your NEXT project...all things considered, it makes analysis, and more importantly practical decision making a lot more efficient, and usually with far less risk all around.