cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
YanivD
Level III

DSD - one of the responses has 2 values 0 or 1 - how to analyze that

Hi,

 

I did a DOE DSD, one of the responses has 2 values - 0 or 1. how to analyze this response - as usual ?

 

 

3 REPLIES 3
P_Bartell
Level VIII

Re: DSD - one of the responses has 2 values 0 or 1 - how to analyze that

"...as usual..." depends on information you didn't share:

 

1. What is your definition of 'usual'?

2. Are the zero and one, true numeric values or a proxy for a categorical response? If the response is a categorical response, then modeling and analysis will take a very different flow compared to a numeric response.

3. If the zero/one response values are numeric, are there possible values between zero and one? Or did the measurement system not discriminate these in between values?

4. Would also help if you shared the design and information regarding the practical problem at hand. Analysis should be aimed at the practical problem at hand. Not too mention the specific design that was used to explore the system.

 

Lots of questions before substantive advice can be given...and I'm sure I've left out a few of those questions...love to hear others thoughts.

YanivD
Level III

Re: DSD - one of the responses has 2 values 0 or 1 - how to analyze that

thanks for your reply. 

i have done an DOE with 3 factors and one of the responses is categorical (0 or 1  i mean categorical).

do i need to define that this response is categorical?

how to do the fit DSD?

 

appreciate your reply:) 

Re: DSD - one of the responses has 2 values 0 or 1 - how to analyze that

No, you should not use Fit Definitive Screening. Instead, change the response modeling type to Nominal and select Analyze > Fit Model. Enter your response data column in the Y role. Select the three data columns for your factors, click Macros, and select Response Surface. Click Run.

 

I am not criticizing your experiment, but three factors is not really a screening situation. It is doubtful that the key screening principles are valid. The success of any screening design, including DSD, depends on these principles. Custom design would be more appropriate in this case. For example, do you intend to use the model to optimize factor levels? Then an I-optimal custom design would perform better than the DSD. I provide this clarification is for future reference.