09-04-2013 08:23 AM
This sounds really kludgy, but the only way I see to do this is to preprocess the data. Plug the fixed value in and subtract it from the dependent variable. This may require using the correct link function. I don't see any other way in GENMOD. The other alternative I can think of would be to do something in NLMIXED that was the equivalent of the GENMOD fitting, and use the BOUNDS statement.
Message was edited by: Steve Denham
10-09-2013 11:10 AM
see the offset statement:for quick and easy "fixing" of parameters to a desired value.
An example here
claim_freq=claim_count / exposure;
if driver_age_group =3 then offset_factor=1.05;
if driver_age_group =4 then offset_factor=1.25;
if driver_age_group in (1,2) then driver_age_group_new= driver_age_group;
else driver_age_group_new =9;
proc genmod data=freq_data;
class driver_age_group_new type;
model claim_freq = driver_age_group_new type
/ dist=poisson link=log offset= logoffset;
10-09-2013 12:36 PM
Matt, how does this fix the parameter value? I can see that it scales it for various age groups, but I was under the impression that the OP wanted a fixed single value, something like:
model y = x1--x10/dist=poi;
with, for example, the contribution of x1, x2, and x3 being constant.