因子篩選實驗是在一個正式實驗前，用來從一大量的因子中篩選出重要的活性因子的先前實驗。在分析因子篩選實驗方面，Modified Box-Meyer Method (MBMM)為根據Box-Meyer 所提出的Box-Meyer Method (BMM)的推廣。在概似函數的假設方面，MBMM用平均數模型取代BMM中的效果模型。此一方法解決了BMM可能面臨的一些最高階交互作用假設錯誤，造成重要因子誤判的問題。此外，解釋變數的個數上也會降低減少了計算上的複雜性。因此MBMM非常的適合用來分析因子篩選實驗。在本文中，我們將利用一次刪除實驗中的一個實驗徑的方式來評估MBMM所辨識出活性因子的穩健性。 Factor screening experiments are previous experiments used to screen for important active factors from a large number of factors prior to a formal experiment. In the analysis of factor screening experiments, the Modified Box-Meyer Method (MBMM) is a promotion of the Box-Meyer Method (BMM) proposed by Box-Meyer. In terms of the hypothesis of the likelihood function, MBMM replaces the effect model in the BMM with the average model. This method solves some of the highest-order interaction hypothesis errors, causing the misjudgment of important factors. In addition, the number of explanatory variables also reduce the computational complexity. Therefore, In this paper, we will use one of the experimental paths in the deletion experiment to evaluate the robustness of the active factors identified by MBMM.
