06-19-2012 10:40 PM
Hi everyone,I was wondering if there is a faster way to do multiple linear regression for multiple outcomes?Even though I can use proc glm for each multiple linear regression (there are about 30 some outcomes, that we will be looking into), it is quite tedious to do so...
example outcomes: SBP, DBP, PP, RASBP, LASBP etc...
independent variables: age, bmi, gender(0, 1), edulevel(1, 2, 3)
outcomes, age, bmi are continuous variablesgender, education levels are categorical
06-20-2012 08:09 AM
PROC GLM can handle all of the outcomes in a single pass, plus it is one of the multithreaded procs, so if you have multiple CPUs available, it can really speed some of the computations. You could try::
proc glm data=yourdata;
class gender edulevel;
model sbp dbp rasbp lasbp <put as many here as you have>=gender edulevel gender*edulevel age bmi;
<insert other stuff here to get lsmeans, etc.>
Let us know if this needs more attention.