Search the Elsmar Cove!
**Search ALL of Elsmar.com** with DuckDuckGo including content not in the forum - Search results with No ads.

Why I get a different kind of regression equation when I do the same data in a differ

#1
I was wondering why I get a different kind of regression equation when I do the same data in a different way(create DOE in the minitab vs type the data myself ).
For example, when I do the job as the picture 1 below, I can get the regression equation:
(If there is no picture, please see the attachment)

However, If I create a new factorial design as the picture 2 below, I got a different equation, and I did not know how to use that equation to predict the response. I did the setting because I would like to analyze the 3 level experiment. I would like to know what the second equation means.
 

Attachments

Last edited:

Miner

Forum Moderator
Staff member
Admin
#2
Re: Why I get a different kind of regression equation when I do the same data in a di

The first equation shows the coefficients for uncoded (natural) units. The second equation shows the coefficients for coded (-1, 1) units.
 
#3
Re: Why I get a different kind of regression equation when I do the same data in a di

I do not think so. In the first one, there is a coded equation coeff. on top of my circle.

I think in the first equation, each variable is treated as continuous. The weird equation may mean that if the first level is present (say power of 50), I put 1 for the power_50 variable and 0 for the power250 variable, etc.

How can I let the second method to treat each variable as continuous so that I can get the clean equation as excel one did?

I am asking because in the second weird equation, I cannot simply put a middle power level(say 150, or the coding value) in the equation to predict the response.
I also want to do the 2 factor 3 level fitting. As far as I know, I can only use the second method and get the weird equation
 

Miner

Forum Moderator
Staff member
Admin
#4
Re: Why I get a different kind of regression equation when I do the same data in a di

I understand. The reason is that you created a General Full Factorial design instead of a 2^k design. General Full Factorials are treated as if the factors have discrete levels rather than continuous.
 
#5
Re: Why I get a different kind of regression equation when I do the same data in a di

Thanks a lot.

So, How do I create 3^k design, not General Full Factorial design, which treated as the factors have continuous levels? I think there is no 3^k option in the window.
 

Miner

Forum Moderator
Staff member
Admin
#6
Re: Why I get a different kind of regression equation when I do the same data in a di

You only have 2 levels, so why would you do that?
 
#7
Re: Why I get a different kind of regression equation when I do the same data in a di

That is just my example to ask the question.
I have a 3 level factorial experiment now.
Therefore, I would like to have a continuous equation for that.
 

Miner

Forum Moderator
Staff member
Admin
#8
Re: Why I get a different kind of regression equation when I do the same data in a di

I would analyze the resulting experiment as a General Linear Model (GLM). You can find this under the ANOVA menu in Minitab.
 
Top Bottom