turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- SAS Programming
- /
- General Programming
- /
- Simulating of Beta-Binomial distribution

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

08-31-2016 12:57 AM

Hi,

I would highly appreciate it if you could give me any guidance how I can simulate the joint Beta-binomial distribution.

Thank you in advance.

Regards,

Yuliya

Accepted Solutions

Solution

09-05-2016
05:16 AM

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Ksharp

09-01-2016 02:30 PM

Hi everyone,

thank you very much for your help.

I simulated with both ways.

Have a nice weekend.

Regards,

Yuliya

All Replies

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Yuliya

08-31-2016 01:27 AM

See if this pdf help you

https://www.sas.com/storefront/aux/en/spsimulationofdata/65378_excerpt.pdf

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to RahulG

08-31-2016 03:30 AM

Hi,

thank you very much. Unfortunately, there is only ine part of the book with the simplest simulations. I need to simulate a joint distribution Beta - binomial. Any ideas?

Thank you again.

Regards,

Yuliya

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Yuliya

08-31-2016 09:40 AM

I think @RahulG's intent was to point you to the book Simulating Data with SAS, rather than suggest that your answer was found in the free chapter.

If you read and understand the blog the @Ksharp linked to, you should have no problem modifying it to the beta-binomial. Start there, and post again if you have a particular question.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Yuliya

08-31-2016 03:38 AM

You should post it at IML forum. Here is Gamma+Possion distribution: http://blogs.sas.com/content/iml/2014/04/02/interpret-nb-distribution.html

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Ksharp

08-31-2016 04:01 AM

Thank you very much!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Yuliya

08-31-2016 11:08 AM

OK. Here is Beta-Binomial distribution. Beta-Binomial is usually for Overdispersion Models . data have; pi=0.3; Rho=0.2; m=30; c=(1-Rho**2)/(Rho**2) ; a=c*pi; b=c*(1-pi); call streaminit(1234); do i=1 to 1000000; p=rand('beta',a,b); x=rand('binomial',p,m); output; end; keep x; run; proc sgplot data=have; vbar x / stat=percent; run;

Solution

09-05-2016
05:16 AM

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Ksharp

09-01-2016 02:30 PM

Hi everyone,

thank you very much for your help.

I simulated with both ways.

Have a nice weekend.

Regards,

Yuliya

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Ksharp

10-26-2017 12:02 AM

Hello Ksharp;

I could not find PDF, CDF, and inverse CDF call functions for beta binomial in SAS. I appreciate your help in determining all possible values of unknown parameters, say eta1 and eta2 (reparameterized parameters) that satisfies two inverse beta binomial functions simultaneously. For example, if the inverse CDF of beta binomial is Q, then how we can solve the following equations together for eta1 and eta2 :

6=Q(0.8, eta1,eta2) and 2=Q(0.1,eta1,eta2) and once we have these values, how we can keep them to be used in a following step in the program.

Thank you

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Abdelkarim

10-26-2017 08:23 AM

Sorry. It is out of my knowledge. But maybe @Rick_SAS could help you. Post your question at IML forum. Since it is about data simulation.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Abdelkarim

10-26-2017 10:44 AM

This thread is old and SOLVED (closed). It contains a link to a blog post that describes how to simulate from a compound distribution. The beta-binomial distribution is compound, so to generate random draws from the beta-binomial you can first draw p from a beta distribution and then draw X from the binomial(p) distribution.

If you want to use the other probability function such as PDF, CDF, and QUANTILE, please start a new thread in in this forum or in the Base SAS Community.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Rick_SAS

10-31-2017 10:43 PM

Hi,

How can I start a new thread in in this forum?

Thank you

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Abdelkarim

11-01-2017 05:40 AM

Click the "Please ask a new question" link at the top right. You can also go up one level to the General SAS Programming Community. However, if it is a question about the beta-binomial distribution you might want to post to the SAS Statistical Community

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Ksharp

4 weeks ago

I'm not sure, but if c=a+b and rho=1/(a+b+1), c=(1-rho)/rho?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Yuliya

4 weeks ago

See the article "Simulate data from the beta-binomial distribution in SAS."

If you also need the PDF, CDF, and quantile functions, see the article "Compute the CDF and quantiles of discrete distributions."