topic Re: How to score a new dataset using estimated parameters from PROC REG in SAS Procedures

How to score a new dataset using estimated parameters from PROC REG

benjamin_2018 — Thu, 30 Jan 2025 12:42:29 GMT

Hello! I am using PROC REG to estimate parameters, and I want to use the estimated coefficients for significant (p<.05) variables to score a new dataset for observations which are missing the modeled variable. I can output the parameters using the outest= option but I don't know how to easily apply these parameters to the dataset which has known values for the independent variables but unknown values for the dependent variable. For example,

ods output SelParmEst=reg_parms_&modeled_ind_var;
proc reg data=work.have;
&modeled_ind_var : model &modeled_ind_var = x y z/selection=backward sls=.05;
run;

I see a reference to a SCORE statement for PROC REG, but I must not have the right version of SAS?

I can see this process done in 1 step using PROC DISCRIM, as below.

%let discrim_var=y;
data missing_&discrim_var not_missing_&discrim_var;
set work.have;
if missing(&discrim_var) then output missing_&discrim_var;
else output not_missing_&discrim_var;
run;

proc discrim data=work.have  testdata=work.missing_&discrim_var testout=results_&discrim_var;
.......;
run;

Is there another procedure that I should be applying than PROC REG?

Re: How to score a new dataset using estimated parameters from PROC REG

sbxkoenk — Thu, 30 Jan 2025 12:54:54 GMT

Techniques for scoring a regression model in SAS
By Rick Wicklin on The DO Loop February 19, 2014
https://blogs.sas.com/content/iml/2014/02/19/scoring-a-regression-model-in-sas.html
The missing value trick for scoring a regression model
By Rick Wicklin on The DO Loop February 17, 2014
https://blogs.sas.com/content/iml/2014/02/17/the-missing-value-trick-for-scoring-a-regression-model.html
4 reasons to use PROC PLM for linear regression models in SAS
By Rick Wicklin on The DO Loop February 11, 2019
https://blogs.sas.com/content/iml/2019/02/11/proc-plm-regression-models-sas.html

BR, Koen

Re: How to score a new dataset using estimated parameters from PROC REG

Tom — Thu, 30 Jan 2025 15:02:29 GMT

Just use the normal OUTEST= option, like it shows in the examples for PROC SCORE.

proc reg data=work.have outest=reg_parms_&modeled_ind_var ;
&modeled_ind_var : model &modeled_ind_var = x y z/selection=backward sls=.05;
run;

Then use that dataset as the SCORE= dataset in your PROC SCORE call.

Re: How to score a new dataset using estimated parameters from PROC REG

benjamin_2018 — Thu, 30 Jan 2025 16:07:47 GMT

Thank you!