proc glimmix

[zyf]

Thanks for your help. I got the error on running the code below. 

ods listing close;
ods output SolutionR=&libbre..mixed_sol_&col._p;
proc glimmix data=&libbre..prob_preds_&col._p method=laplace initglm NOCLPRINT;
nloptions tech=nrridg;
class cap_c_prv_clin_group cap_c_prv_npi;
model had_use (event='yes') = pred_prob / dist=binary solution ddfm=bw;
random int /type=cs subject = cap_c_prv_clin_group s cl;
random int /type=cs subject = cap_c_prv_npi s cl ;
output out = &libbre..mixed_prob_&col._p pred(blup noilink) = pred_yesbl_noil;
run;
ods listing;

ods listing close;
ods output SolutionR=&libbre..mixed_sol_&col._prac;
proc glimmix data=&libbre..prob_preds_&col._prac method=laplace initglm NOCLPRINT;
nloptions tech=nrridg;
class cap_c_prv_clin_group cap_c_prv_npi;
model had_use (event='yes') = pred_prob / dist=binary solution ddfm=bw;
random int /type=cs subject = cap_c_prv_clin_group s cl;
random int /type=cs subject = cap_c_prv_npi s cl ;
output out = &libbre..mixed_prob_&col._prac pred(blup noilink) = pred_yesbl_noil;
run;
ods listing;

NOTE: The GLIMMIX procedure is modeling the probability that had_use='yes'.
ERROR: NRRIDG Optimization cannot be completed.
NOTE: Optimization routine cannot improve the function value.
WARNING: Output 'SolutionR' was not created. Make sure that the output object name, label, or path is spelled correctly. Also, verify that the appropriate procedure options are used to produce the requested output object. For example, verify that the
NOPRINT option is not used.
NOTE: The data set CAP.MIXED_PROB_CAP_LAB_DIFF_P has 73818 observations and 7 variables.
NOTE: Compressing data set CAP.MIXED_PROB_CAP_LAB_DIFF_P decreased size by 90.57 percent.
Compressed is 85 pages; un-compressed would require 901 pages.

NOTE: The GLIMMIX procedure is modeling the probability that had_use='yes'.
ERROR: NRRIDG Optimization cannot be completed.
NOTE: Optimization routine cannot improve the function value.
WARNING: Output 'SolutionR' was not created. Make sure that the output object name, label, or path is spelled correctly. Also, verify that the appropriate procedure options are used to produce the requested output object. For example, verify that the
NOPRINT option is not used.
NOTE: The data set CAP.MIXED_PROB_CAP_LAB_DIFF_PRAC has 73818 observations and 7 variables.
NOTE: Compressing data set CAP.MIXED_PROB_CAP_LAB_DIFF_PRAC decreased size by 90.57 percent.
Compressed is 85 pages; un-compressed would require 901 pages.

Thanks,

Irene

[PaigeMiller]

Optimization methods are not guaranteed to converge at a global maximum, and in fact may not converge at all. There can be many reasons for this, including (but not limited to):

• overspecified model
• sparse data
• bad starting values
• outliers

In your case, where you have two class variables, if they have a lot of levels, you wind up with very sparse data.