Many thanks to Jim Garrett and Jonathan Monteleone for their responses to my
question regarding response surface. The following is the summary :
If you want software that helps you create the experimental design, it
appears that you need S+Dox (or some other package unrelated to S, such as
Minitab). On the other hand, it's actually not too hard to set up a
response-surface design by hand.
If you already have data and you want to fit a model, S+Dox makes it very
convenient but is not required. The model is actually a regression model.
You could create the required predictor variables for the regression and
fit the regression model. For example, if you have predictor variables X
and W and a vector of block labels B, make B into a factor and fit
response.model <- lm(Y ~ B + X + X^2 + W + W^2 + X * W, data = <name of
data frame here>)
You can then do sorts of things, such as check residuals and check the
statistical significance of model components. You can also make
predictions with this model at locations spaced over a grid, and use this
to make a perspective or contour plot. These things require some typing,
so S+Dox would certainly make your life easier, but is not absolutely
required.
Brenda
-----Original Message-----
From: byang@astrixsoftware.com [mailto:byang@astrixsoftware.com]
Sent: Friday, September 14, 2001 3:15 PM
To: s-news@lists.biostat.wustl.edu
Subject: [S] Response Surface
Dear S-Plus users,
Do I need Dox module to analyze Response Surface design? How would I
specify Number of Blocks, Number of Center Points and Number of Runs?
Thanks in advance,
Regards,
Brenda Yang
---------------------------------------------------------------------
This message was distributed by s-news@lists.biostat.wustl.edu. To
unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with
the BODY of the message: unsubscribe s-news
|