MATLAB Answers

where is the pvalue of the model stored in fitlm or fitglm?

109 views (last 30 days)
Theo
Theo on 26 Oct 2014
Commented: Adam Danz on 15 May 2019
I'm looking for the model pvalue in comparison to the constant interception, not the variable pvalues. Thanks

  0 Comments

Sign in to comment.

Answers (4)

Tom Lane
Tom Lane on 27 Oct 2014
Try this:
load hald
lm = fitlm(ingredients,heat)
lm.Coefficients.pValue(1)

  1 Comment

Theo
Theo on 28 Oct 2014
Thanks but that's not what I want. That's the pvalue of the intercept or basically each of the predictors. I'm looking for the model pvalue in comparison to the constant intercept. in your example for instance it's F-statistic vs. constant model: 111, p-value = 4.76e-07

Sign in to comment.


Duijnhouwer
Duijnhouwer on 25 Mar 2016
This works (Matlab 2015b):
load hald
M=fitlm(ingredients,heat)
T=anova(M,'summary')
F=table2array(T(2,4))
pValue=table2array(T(2,5))

  0 Comments

Sign in to comment.


yanarof foranay
yanarof foranay on 28 Sep 2018
This answer may come a bit late, but at least it can help people that google the same problem (like me):
The p-value of the F-statistic vs. constant model (for the fitglm) can be retrieved like this:
pVal = RL_Model.devianceTest.pValue;

  0 Comments

Sign in to comment.


Dameng Yin
Dameng Yin on 15 Nov 2018
Edited: Dameng Yin on 15 Nov 2018
I had the same question. While looking for the answer online, I found the answer to this from here: StackOverflow.
This would work for fitlm as well:
p = coefTest(mdl);
I'm using Matlab 2018a. Not sure if the function is available in previous versions.
Best.

  1 Comment

Adam Danz
Adam Danz on 15 May 2019
And to get the f-statistic too,
[p,f] = coefTest(mdl);

Sign in to comment.

Sign in to answer this question.