topic Re: simulation from negative binomail distribution in SAS Procedures

simulation from negative binomail distribution

Peaw — Mon, 26 Jun 2017 07:16:55 GMT

I am using base sas 9.4. Now I am working on Simulating data.

After I fitted data to negative binomial distribution using the following code ;

proc countreg data=claim;
         model y = / dist=negbin;
         output out=exp_nb prob=prob_nb;
         freq freq;
         run;

     proc format;
         value yfmt
           7-high = ">=7";
         run;

      proc means sum nway data=exp_nb;
         class y;
         var prob_nb;
         format y yfmt.;
         output out=exp_nb sum=_testp_;
         run;

      data exp_nb;
         set exp_nb;
         sumexp + _testp_;
         sumtolast = lag(sumexp);
         if y=7 then _testp_ = 1 - sumtolast;
         run;

      proc freq data=claims;
         table y / chisq(testp=exp_nb df=-2 lrchisq);
         format y yfmt.;
         weight freq;
         run;

Then I got only p-value from fitting this data.

Now I want to simulate data from negative binomial distribution.

How can I do that ? please help

(version of sas program do not have Proc iml procedure)

Re: simulation from negative binomail distribution

PeterClemmensen — Mon, 26 Jun 2017 06:31:41 GMT

You can simulate data from a negative binomial distribution using the RAND function like this

data negbin;
	do x = 1 to 1000;
		y = rand('negbinomial', 0.5, 2);
		output;
	end;
run;

title 'Plot the Simulated Data';
proc sgplot data = negbin;
	histogram y;
run;
title;

Adjust p and k to your liking. The documentation is here

http://support.sas.com/documentation/cdl/en/lrdict/64316/HTML/default/viewer.htm#a001466748.htm

Re: simulation from negative binomail distribution

Peaw — Mon, 26 Jun 2017 07:23:32 GMT

could you please tell me that how can I know p of this data from fitting the distribution.

because when I fitted this distribution I fitted from raw data of frequency of claim of each auto policy. That's why I do not know P of this data.

Re: simulation from negative binomail distribution

Ksharp — Mon, 26 Jun 2017 14:18:10 GMT

Ha. Here is.

http://blogs.sas.com/content/iml/2012/04/04/fitting-a-poisson-distribution-to-data-in-sas.html

proc genmod data=MyData;
   model N = / dist=negbin;
   output out=Fit p=lambda;
run;

http://support.sas.com/kb/24/166.html

Re: simulation from negative binomail distribution

Peaw — Thu, 29 Jun 2017 03:30:23 GMT

I am sorry to bother you. But after following your suggestion I got the attached output.

How can I know this model fits the data or not.

Re: simulation from negative binomail distribution

Ksharp — Thu, 29 Jun 2017 12:29:37 GMT

 It looks good.  
Value/DF is near  1 .