Count Data

[vdfdd]

Please I need suggestions on how to simulate count data from a Poisson distribution and fit (using PROC GLIMMIX) Normal, lognormal and negbin to the resulting data.Thank you

[Reeza]

https://www.amazon.ca/Simulating-Data-SAS-Rick-Wicklin-ebook/dp/B00CC0DQ16

[vdfdd]

Thank you for your swift response.

[StatDave]

Use the RAND('POISSON', mean) function in the DATA step to generate Poisson distributed data. Just set the desired value of the Poisson mean to produce a set of values. For example, these statements produce 100 observations in which Y has Poisson distributed values with mean=2.

data s;
do i=1 to 100;
y=rand("poisson",2);
output;
end;
run;

[PeterClemmensen]

Are you sure PROC GLIMMIX is the right tool for fitting this data?

[vdfdd]

I am not completely sure. I was looking at the example in this link: https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_glimmix_sec...

[Reeza]

What are you trying to do overall? Why the need to simulate data?

---

@vdfdd wrote:

I am not completely sure. I was looking at the example in this link: https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_glimmix_sec...

---

[vdfdd]

To test which distribution best fit the (non-normal, discrete) data

[Reeza]

PROC UNIVARIATE can test this with histogram statements. 

[vdfdd]

Thank you for your suggestion.

[Ksharp]

You could find answer at  @Rick_SAS  blog.

http://blogs.sas.com/content/iml/