RANUNI function assigns the same number to 90% of records

[HeidiDT]

Hi All

I am hoping someone has come across this issue previously and knows why it happens: I am assigning a random number between 1 and 1,000 to a daily feed of data by making use of the RANUNI function:

                SAS_RANDOM_NUMBER  = int(1000*ranuni(-1));

The problem I have is that on certain days the same number is allocated to 90% of the records, so not so random after all. Has anyone ever come across this problem? 

[PeterClemmensen]

Try using the Rand Function instead. 

Rick Wicklin explains why in the article Six reasons you should stop using the RANUNI function to generate random numbers.

[Kurt_Bremser]

Don't use RANUNI, use the improved RAND:

call streaminit(-1);
SAS_RANDOM_NUMBER  = int(1000*rand('uniform'));

If you want a uniform integer distribution, use

rand('integer',1000);

[HeidiDT]

Thanks guys! Will that explain why more than 90% of the records get the same number assigned though? Is RANUNI that "bad"?

[Kurt_Bremser]

Well, I never experienced something as egregious as what you describe, but then I rarely needed random numbers. I do mostly ETL work, no simulations.

And I've been using RAND since it was introduced.

[Watts]

It looks like you're specifying a seed value of -1, which uses the time of day to initialize the seed stream. I don't know how you're using the RANUNI function to generate the list of random numbers, but if (?) you're including the function in a loop, it's possible that this runs faster than the clock updates -- and therefore reuses the same seed.

For more information, see How to Generate Random Numbers in SAS by @Rick_SAS .

[Rick_SAS]

If you post the code you are using, we can say more about what you are seeing and why.