论文部分内容阅读
Varied designs have disparate constructions and different confounding situations.Most models focus on the description of the whole confounding situation of a design.However,actually different columns have different confounding situations.Aware of the situation,practitioners can arrange factors more wisely.Zhou,Balakrishnan and Zhang (2012) proposed a new pattern called factor aliased effect number pattern (F-AENP for short) to depict confounding situation of every column in a GMC design and rank the columns.Using this new pattern,we have calculated F-AENP of all the regular designs of 2n-m with 16 runs and 32 runs.By F-AENP,all the columns in a design can be ranked into a new order.Further more,we obtain the compromise designs applying to the condition that one or two very important factors and two factor interaction involving them should be estimated,and simultaneously experimenters hope them for having confounding situation as clear as possible.