topic Re: R - Square in GLM in Statistical Procedures

R - Square in GLM

JCW91 — Wed, 23 Aug 2017 03:28:35 GMT

After running GLM, the statistics that showns are Deviance, AIC, AICC, BIC etc.... without R-square generated. How do i get the R- square everytime I have run the GLM?

* I notice that there is a formula for the R-squared but I try to avoid it computing manually.

Thanks in advance.

Re: R - Square in GLM

Reeza — Wed, 23 Aug 2017 04:18:38 GMT

What does your code look like that doesn't generate an R Squared? Does it include categorical variables?

Re: R - Square in GLM

Rick_SAS — Wed, 23 Aug 2017 12:33:43 GMT

The R-Square value is in the FitStatistics table, which is created automatically. Here are two examples where I've use ODS SELECT to display ONLY the table that has the R-square value

proc glm data=sashelp.class plots=none;
model height = weight;
ods select FitStatistics;
quit;

proc glm data=sashelp.class plots=none;
class sex;
model height = sex;
ods select FitStatistics;
quit;

Re: R - Square in GLM

mh2t — Wed, 24 Jun 2020 15:35:15 GMT

@Rick_SAS how can we get an output in GLM for multiple variables like below:

VarName RSQUARE height_MEAN
weight xx xxx
sex yy yyy

Thank you for your response!

Re: R - Square in GLM

Rick_SAS — Wed, 24 Jun 2020 18:01:14 GMT

You can get the RSquare values for the models by putting

ods output FitStatistics(PERSIST)=Results; /* concatenate into data set named 'Results' */

prior to the first PROC GLM call.

The mean response does not change from model to model, so that will be a constant column.