- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
DOE doubts
I have 4 different factors in my experiment, each of the factors have around 3 to 5 levels, however, while making DoE using customized design, the table is taking up only higher levels and the lower levels are omitted from combinations. I do not want to do a full factorial design as the number of runs would be very high but in customized design, lower values are omitted, can you please help tell how to obtain a table where lower values are not omitted or is there an option where I can specify the values to be included which are mandatory for my experiment.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: DOE doubts
Thanks for your request. There is some information missing to give a definitive answer, but you can use the discrete numeric factor option in case your factors are continuous. There you can specify specific factor levels for that factor. In your case you should also add higher order terms to your model in the custom designer as well. How does your model specification looks like?
Hard to tell, why the custom designer is omitting the lower levels in your design based on your description. Could you provide an example of your specifications in the custom designer?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: DOE doubts
Hi @Rachana1,
Welcome in the Community !
May I suggest you take a look at similar post and responses here : https://community.jmp.com/t5/Discussions/force-levels-in-DoE/m-p/751878#M93311
As suggested by @Jonas_Rinne, you can specify your factors as discrete numeric with a specific number of levels to force the generation of the needed levels in your design (JMP automatically adds the necessary higher order terms in the model, and you can switch back the factors formats to continuous numeric later for the analysis for example), or keep the factors as numeric continuous and simply specify the higher order terms needed in the model to create equidistant levels in the design : up to 3rd order term (for example X1, X1.X1 and X1.X1.X1) in the model to have 4 equidistant levels in the design (for X1, coded as -1, -0.33, 0.33 and 1).
Hope this answer will help you,
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: DOE doubts
Thank you so much for the link.