MATLAB Answers

laurie
0

LinearModel.stepwise Matlab R2011a

Asked by laurie
on 4 Apr 2012
Hi there
I am trying to perform a stepwise regression. The code i tired is the following :
mdl = LinearModel.stepwise(X,y,modelspec,'VarNames', {labx1,labx2,labx3,labx4,laby});
and i got te following error message :
??? Undefined variable "LinearModel" or class
"LinearModel.stepwise".
it is weird because i used the exact same code as the one described here :
I am afraid this is because i have an older release of Matlab. I use R2011a with Statistics Toolbox 7.5. I have been trying to read the "documentation for older releases" but the website would accept neither my License number nor my activation key so i can't access the documentation. This is a student license.
What can I do ? Thank you very much Laurie

  0 Comments

Sign in to comment.

2 Answers

Answer by the cyclist
on 4 Apr 2012
 Accepted Answer

You are correct to guess that the problem is that you have an older release. However, you still have some options in R2011a. There are two functions for stepwise regression: stepwise() and stepwisefit(). One of them uses an interactive UI, and the other does not.

  0 Comments

Sign in to comment.


Answer by laurie
on 4 Apr 2012

Hi cyclist :)
Thank you ! that is exactly what i was looking for. Do you know if stepwisefit() automatically includes interactions ? or do i have to add those regressors manually ? apparently yes but this seems surprising ?

  2 Comments

I would check "doc stepwisefit" to be 100% certain, but I'm pretty sure you have to add interactions terms separately.
For future reference, your "answer" here would have been better placed as a comment on my answer. Also, please consider accepting my answer, which might help future users with a similar question.
laurie
on 4 Apr 2012
okay thank you :)
ps : i found that X=x2xf(X,'interaction') allows one to add the interactions in X so you don't have to construct them manually. you also have to delete the "constant" column that x2xf adds as the first column of X and that stepwisefit() doesn't need.

Sign in to comment.