09-06-2017 12:04 PM
For the attached sample dataset, I would like to:
1) generate a variable "random_value_norepeat" which equals to a randomly selected number from the variable "value" (any number from the "value" can only be selected for one time, or put it another way, sample without replacement), and
2) generate another variable "random_value_repeat" which also equals to a randomly selected number from the variable "value", but the number from the "value" can be selected more than once (sample with replacement).
09-06-2017 01:05 PM
Google turns up lots of examples
09-06-2017 01:10 PM
proc surveyselect does both. method=srs selects without replacement, method=urs does so with replacement.
09-06-2017 01:12 PM
All the information you need is at
I believe you have to generate each variable separately. It is not clear if you want all 50 observations or a subset. It is also not clear if you need the values in random order or in the original order. The easiest way is to use the SAMPLE function in SAS/IML, but since you posted to the Base SAS Community, here is a PROC-based solution to get you started:
libname temp "C:\Temp"; data Have; set temp.have; call streaminit(1); u = rand("uniform"); run; /* without replacement (SRS) */ proc sort data=Have out=SRS(rename=(value=random_value_norepeat)); by u; run; /* with replacement (URS) */ proc surveyselect noprint data=temp.Have method=urs seed=2 samprate=1 outhits out=URS(rename=(value=vandom_value_repeat)); run; /* if you want URS in random order, then generate random uniform and sort */