logistic regression for model prediction

momolito — Thu, 22 Feb 2024 03:57:01 GMT

Hello

I am trying to develop a prediction model for oral cancer using logistic regression using the code below. the final model does not include some known risk factors for oral cancer. How can I modify the code below, to force DRKG, SMOKE, and BNUT among the final selected predictors? Thanks

proc logistic data=SETA;
class sex (REF='1') AGE_GROUP (REF='1') BMI_GRP (REF='1') SMOKE (REF='0') DRKG (REF='0') BNUT (REF='0') EXERCISE (REF='0') MAR_STAT (REF='1') INCOME_GROUP (REF='1') FAMHIST (REF='0') RESIDENCE (REF='1') EDUC (REF='1') MOM (REF='1') FA (REF='1');
model Oral(event='1') = Smoke DRKG BNUT AGE_GROUP SEX BMI_GRP EXERCISE MAR_STAT INCOME_GROUP RESIDENCE EDUC MOM FA FAMHIST/ selection=STEPWISE slstay=0.05;
output out=oral_cancer_final p=predicted;
run;

Re: logistic regression for model prediction

StatDave — Thu, 22 Feb 2024 04:24:16 GMT

Add the INCLUDE=3 option in the MODEL statement. Be sure to keep those three variables as the first ones listed after the equal sign.

Re: logistic regression for model prediction

momolito — Thu, 22 Feb 2024 04:29:16 GMT

It worked. Thanks alot for the quick help

topic logistic regression for model prediction in Statistical Procedures

logistic regression for model prediction

Re: logistic regression for model prediction

Re: logistic regression for model prediction