Jen1978
09-26-2007, 12:01 PM
Has anyone ever done bootstrapping on a dataset for which you were planning to do a regression analysis? My dataset is pretty small so I'd like to ensure stability of the predictors in my model but I'm not sure about how to write the code for this. What I'm hoping is to run about 100 different bootstrap samples of 1000 observations each, do univariate regression on each for the variables 'age', 'Charlson', 'warfarin' and 'sex1' and then do a logistic regression on the variables that are significant based on the bootstrapping (using 'outcome' as the dependent variable)
here's the code so far (without bootstrapping):
proc logistic data=miout;
model outcome (event='1')=age;
run;
proc logistic data=miout;
model outcome (event='1')=Charlson;
run;
proc logistic data=miout;
model outcome (event='1')=warfarin;
run;
proc logistic data=miout;
model outcome (event = '1') =sex1;
run;
proc logistic data =miout;
model outcome (event = '1') = age Charlson warfarin sex1/;
run;
If anyone has any idea how to incorporate bootstrapping code, I'd REALLY appreciate it...I'm at a total loss!
Thanks very much!
here's the code so far (without bootstrapping):
proc logistic data=miout;
model outcome (event='1')=age;
run;
proc logistic data=miout;
model outcome (event='1')=Charlson;
run;
proc logistic data=miout;
model outcome (event='1')=warfarin;
run;
proc logistic data=miout;
model outcome (event = '1') =sex1;
run;
proc logistic data =miout;
model outcome (event = '1') = age Charlson warfarin sex1/;
run;
If anyone has any idea how to incorporate bootstrapping code, I'd REALLY appreciate it...I'm at a total loss!
Thanks very much!