Jan 21, 2015
I have a DOE design where one of my 20 factors has a low probability in real life of being "on".
When the factor is "on", it has a very dominant effect on most outputs.
Normal DOE design would split the factor state 50/50 on/off across all tests.
It would seem a better DOE design where less than 50% of tests have this factor set to "on".
Is there a standard method I would somehow bias factor state to "off", the more typical, lower dominant state? Thanks - Paul