Wally,
I’d tread lightly regarding Taguchi methods of experimentation. It has been shown in numerous periodicals and statistical texts that there are some fundamental flaws in this method of experimentation. One of the major issues revolves around the methodologies lack of power in being able to accuratelt detect and predict the factors that impact both the y=f(x) and s=g(x).
If you have a choice use the factorial approach to experimentation. If a primary reason for interest is the use of the inner and outer array, then look to the use of “chunk variables” to set several noise and/or environmental settings at appropriate high and low levels. Good luck with your analysis.
Regards,
Erik
.
