ERROR: Floating Point Zero Divide PROC GENMOD

AudreyM — Mon, 19 Dec 2022 11:43:58 GMT

i have to do this proc genmod:

PROC GENMOD DATA = _1;
CLASS TRTP1-TRTP3 country1-country21 NEXTGR11-NEXTGR12 FEVPGR11-FEVPGR13;
MODEL AVAL = TRTP1-TRTP3 country1-country21 NEXTGR11-NEXTGR12 FEVPGR11-FEVPGR13 /
OFFSET=LNFUDUR2 DIST=negbin LINK=log;
BAYES SEED=22022021 COEFFPRIOR=uniform THIN=100 OUTPOST= dataset2 NMC=100000;
RUN;

but unfortunately, i have the below error messages (all the covariates are binary 0/1 and the AVAL variable includes only integer values from 0 to 5. Only the OFFSET variable is a continuous variable)

The distribution of AVAL is the following:

my thought is that is due to the low variability of AVAL because in a lot of strata, the available values of AVAL are only 0.

But these are the data that i have so i cannot change them. i have tried to round the offset variable so have a fix number of digits but does not change the outcome.

Would you have any solution to resolve this?

Thanks

Audrey

Re: ERROR: Floating Point Zero Divide PROC GENMOD

PaigeMiller — Mon, 19 Dec 2022 13:11:12 GMT

Hessian not positive definite often is a result of the model not fitting the data well. See https://communities.sas.com/t5/Statistical-Procedures/mixed-model-NOTE-Convergence-criteria-met-but-final-hessian-is/m-p/494284#M25630

topic ERROR: Floating Point Zero Divide PROC GENMOD in SAS Studio

ERROR: Floating Point Zero Divide PROC GENMOD

Re: ERROR: Floating Point Zero Divide PROC GENMOD