| Summary: | 碩士 === 國立中興大學 === 統計學研究所 === 105 === In industrial experiments, not all factors can be randomly and independently reset at each run. There are many constraints in practice and the levels of some factors is not easy to change. Some designs were developed to solve this problem, including the split-plot design, split-split-plot design, and staggered-level design. In this paper, we discuss the staggered-level design, which allows two types of factors to reset their levels at different time. We use the optimal D criterion and the Alias-optimality to find designs which can maintain high D efficiency and reduce the influence of potential effects on the primary effects in the model.
|
|---|
