Deciding suitable model from DoE

nac

Dear All,

I created a definitive screening design with 1 factor categorical and 3 factors continuous. 14 experiments obtained from definitive screening design part. I also performed two experiments additionally 14 experiments obtained from definitive screening. You can see additional studies with green highlighted at the attached jmp file. 

When I planned these studies, I copied DoE plan from jmp to word file. After results obtained, I try to put results on DoE but I do not obtained the same obtaination from 4 factors. I try to make assessment of obtained results by using fit model platform but ı do not understand how can ı make the finding suitable model according to obtained results with this DoE. When I try to use "fit definitive design" platform jmp gives error.

Could you please anyone can help me to find creating suitable model for my desing.

 

Best Regards,

 

Yield predicition from definitive screening desing.jmp

Mark_Bailey

First of all, you should not export the design from the JMP data table and then import it back into another JMP data table. JMP creates a data table when you click Make Table with data columns for factor settings and data columns to store the responses. These data columns also have important meta-data associated with them that are called upon during the analysis. You import the data but the meta-data is lost.

 

Second, the additional runs destroy the special structure of the definitive screening design. The Fit Definitive Screening approach cannot work with the modified experiment. You can exclude the rows for the additional runs and then you can use Fit Definitive Screening to select the model.

 

You can use all the observations with Fit Least Squares or Stepwise to select the model.