topic Re: Creating a new distribution from 3 others in New SAS User

Creating a new distribution from 3 others

iressa1313 — Mon, 01 Oct 2018 17:22:08 GMT

Hi I have used SAS to generate 3 different distributions

1. N~ (0,1) with variable X1.....X20

2. N~ (5,1) with variable Y1.....Y20

3. B(0.5) "bernoulli" with variable U1......U20

Now I want to create a new distribution Q where

Q { X, if U = 0 and Y, if U =1

How would I do this? Would I create a new dataset with all 3 variables?

Thanks for any help!

Re: Creating a new distribution from 3 others

ballardw — Mon, 01 Oct 2018 17:29:24 GMT

@iressa1313 wrote:

Hi I have used SAS to generate 3 different distributions

1. N~ (0,1) with variable X1.....X20

2. N~ (5,1) with variable Y1.....Y20

3. B(0.5) "bernoulli" with variable U1......U20

Now I want to create a new distribution Q where

Q { X, if U = 0 and Y, if U =1

How would I do this? Would I create a new dataset with all 3 variables?

Thanks for any help!

Probably, or merge existing data sets.

When you say X, if U = 0 and Y, if U =1 do you mean if U=0 and U=Y or something else? Be very explicit. Also there appears to be something missing after "If U=1 …"

Re: Creating a new distribution from 3 others

Astounding — Mon, 01 Oct 2018 17:40:40 GMT

You'll need to describe the data sets you have a little more. For example, in the X data set:

Do you have 20 observations, each with a value for the variable X? (If so, is there a separate variable taking on values 1 through 20?)

Do you have 1 observation, with variables named x1 through x20?

Do you have many observations, each with variables named x1 through x20?

It can't be all that difficult to compute Q, but it's a whole lot easier to program it once instead of three times. So describe the input a bit more.

Re: Creating a new distribution from 3 others

FreelanceReinh — Mon, 01 Oct 2018 21:46:22 GMT

Hi @iressa1313,

I would keep it simple and create a new dataset with all four variables:

data want;
call streaminit(27182818);
do _n_=1 to 20;
  x=rand('norm');     /* X ~ N(0,1)     */
  y=rand('norm',5);   /* Y ~ N(5,1)     */
  u=rand('bern',0.5); /* U ~ Bin(1,0.5) */
  q=ifn(u,y,x);       /* normal mixture */
  output;
end;
run;

Or is there something special about your existing "distribution datasets" that you don't want to discard them?