Reputation: 15
I am fitting a mixture model consisting of 3 Poisson distributions. For my report I would like to obtain the histogram of the response together with the fitted distribution. Normally, it is as simple as using plots=density option in the proc fmm statement, so my code is as follows:
ods graphics on;
proc fmm data=hw1 plots=density(bins=30);
model frequency= / dist=poisson k=3 parms(2.21,2.59,3.13);
probmodel / parms(-0.69,0.11);
freq count;
run;
ods graphics off;
My problem is that the code produces only the histogram without the density. I was wondering if anybody knows if it could be related to the Poisson distribution and proc fmm does not support it? Because when I change the dist=poisson to dist=gaussian it works.
Thank you
Upvotes: 1
Views: 289
Reputation: 12465
This is because the Poisson distribution is a discrete and not a continuous distribution. If you need a continuous distribution, try something like the normal or weibull distributions.
Upvotes: 1