Help using Base SAS procedures

Solutions for proc glimmix convergence issues

Reply
N/A
Posts: 1

Solutions for proc glimmix convergence issues

Hi SAS help,

I am creating odds ratios for each individual county using a multilevel technique with proc glimmix procedure. However, I have encountered convergence issue and the sas log showed that "Pseudo-likelihood update fails in outer iteration 1." I am using a very large dataset contains more than 2 millions of records. The cluster size ranged a lot (e.g., from 70 to 150,000). I was successfully using the procedure below to create the odds ratio for each county in our study area for a specific year. However, this procedure didn't work for the dataset in another year. Could anyone have any suggestion about this? Thanks a lot!!

The procedure that I used is as following:

proc glimmix data= s09;

class county09 Sex(ref='1') racegp1(ref='0') racegp2(ref='0');

model deny (event=last)= racegp1  racegp2  Sex  LaonAmount Income/solution dist=binary link=logit oddsratio cl;

random intercept/subject=county09 solution cl;

random racegp1 racegp2/subject=county09 solution cl;

ods output solutionf = estimate_fix  solutionr=estimate_random covparms=zcov  fitstatistics=zfit ;

run;

Yenting

Ask a Question
Discussion stats
  • 0 replies
  • 202 views
  • 0 likes
  • 1 in conversation