If you don't have results for all levels of all dimensions can you still run OptimizeIQ? Could it pick the best from what results exist for or even better recommend a combination that we don’t have results for – like a fractional factorial matrix where you use predicted estimates to fill in the missing cells?
Aug 19, 2014
It will still consider all combinations of levels, and use default values for anything not seen in the data. Probably better to provide it estimates for unseen levels as you say, but will at least use defaults.