topic Re: a simple linear regression model that I need to run repeatedly, hope to have an automated soluti in SAS Programming

a simple linear regression model that I need to run repeatedly, hope to have an automated solution

rykwong — Wed, 11 Jan 2017 18:51:04 GMT

Dear all,

I have this logistic regression model and would like to extract the p value of the interaction term "drug*var1".

proc glm data=test;

model bodyweight = drug var1 drug*var1;

run;

Problem is I need to do this 600 times because there are var1, var2, var3..... .... var600. Any thoughts appreciated!!

Raymond

Re: a simple linear regression model that I need to run repeatedly, hope to have an automated soluti

Karlb — Wed, 11 Jan 2017 21:10:23 GMT

Hi,

the idea is as follows

with

ODS listing off;

you can stop procedure list output

with

ODS output....;

you can direct procedure results (in your case the p values) to datasets, please check exact syntax...

finally you macro loop around the Proc glm

%macro writep;

%do k=1 %to 600;

ods output ...out=pdset&k;

proc glm;

model bodyw = drug var&k drug * var&k;

run;

%end;

proc append base=pvalues data=pdset&k; run;

%mend;

%writep;

hope it helps...

sincerly Karl

Re: a simple linear regression model that I need to run repeatedly, hope to have an automated soluti

Rick_SAS — Mon, 13 Feb 2017 14:19:11 GMT

Although Karlb's solution will work, it tends to be slow. An alternative solution is to transform the data from wide form to long form. The new form of the data will contain 600 BY groups. The first contains the data for Y, Drug, and X1. The second contains the data for Y, Drug, and X2. And so on. For an example and discussion, see the article

"An easy way to run thousands of regressions in SAS"