topic Re: Wright-Fisher simulation in Statistical Procedures

Wright-Fisher simulation

jeffgreen — Thu, 08 Oct 2020 15:39:20 GMT

Hi everyone,
I have created the following dataset:
 
data binomial;
input n p;
cards;
12 .5
9 .4
15 .6
;
run;
 
 
Thanks in advance
 
 

Re: Wright-Fisher simulation

SASJedi — Thu, 08 Oct 2020 13:23:48 GMT

I moved this post to the Statistical Procedures forum, where it should get noticed by folks more familiar with topics like this.

Re: Wright-Fisher simulation

FreelanceReinh — Thu, 08 Oct 2020 14:57:31 GMT

Hi @jeffgreen,

Try this:

data want;
call streaminit(27182818);
set binomial;
k=n*p;
do r=1 by 1 until(k=0 | k=n);
  k=rand('binom',k/n,n);
  output;
end;
run;

If you don't need the individual binomial random numbers in dataset WANT, just omit the OUTPUT statement. Then you'll get only the final random number k (equal to 0 or n) and the number r of random number generations which led to this result.

Re: Wright-Fisher simulation

ballardw — Thu, 08 Oct 2020 15:06:01 GMT

@jeffgreen wrote:

Hi everyone,

I have created the following dataset:

data binomial;
input n p;
cards;
12 .5
9 .4
15 .6
;
run;

Basically, I want to run a Wright-Fisher simulation. For example, in the first line of code, I want to run binomial(12, .5), and using this result (say for example the result was 5), I want to do another binomial calculation- binomial (12, 5/12) where 5/12 is the result of the previous binomial simulation divided by n (12). I want to create code that, for each line of code, will repeat this until the generated number is 0 or n (which means the binomial simulation is either a success 0% of the time or 100% of the time). How would I go about doing this?

Thanks in advance

Please describe in some detail what this phrase means: " I want to run binomial(12, .5)".

SAS provides a number of functions that will let you work with specified distributions like binomial with given parameters: CDF PDF PROBBNML but "run" does not describe much.

Re: Wright-Fisher simulation

jeffgreen — Thu, 08 Oct 2020 15:32:13 GMT

Thank you so much!! Worked perfectly!