topic Re: Add dummy values to column in New SAS User

Add dummy values to column

SASPhile — Thu, 27 Jun 2019 15:29:21 GMT

I have a dataset with 3000 records. One of the variables is blank for now,but we expect values in next refresh. How to add dummy values with no specific order and also the count of each value is not main concern.

for example variable_reqd is blank in dataset "have" . But the variable need to have values 'a','b','c','d' in "want" for all 3000 rows.

Re: Add dummy values to column

Kurt_Bremser — Thu, 27 Jun 2019 15:32:56 GMT

data want;
set have;
array r{4} _temporary_ ('a','b','c','d');
variable_reqd = r{rand('integer',1,4)};
run;

Re: Add dummy values to column

SASPhile — Thu, 27 Jun 2019 16:17:53 GMT

There is an ERROR reported:

ERROR: Attempt to initialize element of numeric array r with
character constant 'a'.

Re: Add dummy values to column

Kurt_Bremser — Thu, 27 Jun 2019 16:22:46 GMT

My bad. Add a dollar sign to the array statement:

array r{4} $ _temporary_ ('a','b','c','d');

Re: Add dummy values to column

SASPhile — Thu, 27 Jun 2019 16:24:35 GMT

I tried that before reaching out. But still the below error is reported:

5553 array r{4} $ _temporary_ ('a','b','c','d');
5554 variable_reqd = r{rand('integer',1,4)};
5555 run;

ERROR: The first argument of the RAND function must be a
character string with a value of BERNOULLI, BETA,
BINOMIAL, CAUCHY, CHISQUARE, ERLANG, EXPONENTIAL, F,
GAMMA, GAUSSIAN, GEOMETRIC, HYPERGEOMETRIC, LOGNORMAL,
NEGB, NORMAL, POISSON, T, TABLE, TRIANGULAR, UNIFORM, or
WEIBULL.

Re: Add dummy values to column

Reeza — Thu, 27 Jun 2019 16:29:47 GMT

You're likely on a version of SAS that doesn't have the integer option for rand yet. What is your SAS version?

Rick Wicklin has a macro on his blog to replicate the function.

Re: Add dummy values to column

SASPhile — Thu, 27 Jun 2019 16:33:11 GMT

SAS version:9.04.01M3P062415

Re: Add dummy values to column

Reeza — Thu, 27 Jun 2019 16:37:24 GMT

array r{0:3} $ _temporary_ ('a','b','c','d');
variable_reqd = r(mod(_n_, 4);

If it doesn't matter where they go, just loop them.

Re: Add dummy values to column

Kurt_Bremser — Thu, 27 Jun 2019 16:50:56 GMT

Then you need to use the uniform distribution with a little additional work:

r{floor(rand('uniform')*4)+1}

Edit: changed normal to uniform

Re: Add dummy values to column

ballardw — Thu, 27 Jun 2019 17:42:09 GMT

@Kurt_Bremser wrote:

Then you need to use the uniform distribution with a little additional work:
r{floor(rand('uniform')*4)+1}
Edit: changed normal to uniform

or r[rand('table',.25,.25,.25,.25)} might work.

I don't remember which version introduced the rand('table').

Re: Add dummy values to column

Reeza — Thu, 27 Jun 2019 17:46:15 GMT

rand table is old, 9.2 or 9.1 I think so would work here.