# Discrete factor in DOE

Six Sigma – iSixSigma Forums Old Forums General Discrete factor in DOE

Viewing 5 posts - 1 through 5 (of 5 total)
• Author
Posts
• #33392

ROSS
Member

What should we do in DOE analyse if the factor is discrete? How can we design/do and analyse? Thank you very much!

0
#90214

Member

Set your settings at -1 (least desirable setting) and 1 (most desirable setting). If you have 3 levels center point is 0. If you run a linear regression the transfer function or regression equation will lead you to correct setting.

0
#90219

Dr. Scott
Participant

Tony,
What sort of discrete data are you faced with? If binomial or poison, you can use univariate tests on these. Also, by multiplying the -1 and +1 settings to create surrogates of interactions, you can ultimately conduct multivariate analyses.
However, this is seldon necessary unless your sample numerator is very small.
Dr. Scott

0
#90223

ROSS
Member

Thank you very much! I will try it accordding to your instruction.
And I am not familiar with the discreat factor analysis in regression /DOE…Can you provide me some materials or samples about it? Thank you very much!
My emaul address:  [email protected]

0
#90242

Dr. Scott
Participant

Tony,
My apologies, I did not read you question closely enough. I was answering to discrete Y variable, not discrete X variables or factors.
With discrete factors or X variables, the analysis is conducted as it normally is. However in Minitab, you must specify that the X variables are discrete by giving them text level settings rather than -1 or +1. For example, Factor A settings might be “low” and “high” rather than -1 and +1, and Factor B might be “method 1” and “method 2”, etc.
Doing so will force the transfer function or “response optimizer” in Minitab to treat the variables as discrete rather than continuous.
Hope this Helps,
Dr. Scott

0
Viewing 5 posts - 1 through 5 (of 5 total)

The forum ‘General’ is closed to new topics and replies.