Fractional Factorial Designs
In the two-level experiment with k input factors, there are 2 k possible runs. For k = 3, there are only eight runs. But even with five factors, 25 = 32 runs may be prohibitively expensive and time consuming. The question is whether it is possible to use some subset of all possible runs and still be able to fit a first order polynomial approximation to the response function. The answer is yes, but you must give up some resolution in your approximation. The amount of resolution lost is generally in terms of which interaction effects which can be included in your model.
KeywordsApply Scientist Fractional Factorial Lower Order Term Generator Generator Screen Shot
- Montgomery, D. C. (2001). Design and analysis of experiments. New York: Wiley.Google Scholar