I am trying to generate predictions for an exponential distance-decay model of a response variable (Y) varying with distance (D), i.e. Y = a*exp(-b*D). My problem is how to generate credible intervals for each predicted Y. I wrote the following code in the GENMOD procedure of SAS.

ods graphics on;

proc genmod data=mydata;

model logY = Distance/ dist = normal;

bayes seed = 4861 NBI=500 nmc = 10000 cprior = jeffreys diagnostics = all summary = all;

output out=newout pred = predict upper = LCL lower = UCL;

Run;

ods graphics off;

Proc print data = newout;

run;

The Bayesian point estimates and their equal-tail and highest density intervals are fine. However the lower and upper limits I got are the maximum likelihood confidence limits. I guess there is a way of tweaking the "output out=newout pred = predict upper = LCL lower = UCL;" command line to get the equal-tail intervals.

With the MCM procedure the "output" command was not recognized.

Any command in R?

More Gudeta Weldesemayat Sileshi's questions See All
Similar questions and discussions