topic concatenating and subsetting data sets in SAS Programming

concatenating and subsetting data sets

SAS_inquisitive — Thu, 17 Dec 2015 22:47:31 GMT

data a;
input x y;
cards;
1 3
4 5
6 7
;
data b;
input x y;
cards;
1 4
4 7
8 9
;

data test;
   set a b;
   if x=1 then delete;
run;

I have a test code above. I want to know the process flow of this program. Does if statment acts separately on data sets i.e. read observation from a and apply if stament on each observation and after finishing on a starts on b?

Re: concatenating and subsetting data sets

ballardw — Thu, 17 Dec 2015 23:08:40 GMT

Unless you specifically tell it to examine values from only one of the data sets then any record with a value of 1 for x will be deleted.

You can use the IN dataset option to identify which dataset contributed the observation if that is needed. The IN option creates temporary variables that can be used for testing membership. The example below says to delete only if the record came from dataset b. The IN for dataset A would be completely optional for this code but provided as an example. If you are MERGEing datasets you can use code like If InA and InB ... to see if the record has contributions from both (or more than two) datasets.

data test;
   set a (in= inA)
       b (in= inB);
   if inB AND x=1 then delete;
run;

Re: concatenating and subsetting data sets

SAS_inquisitive — Thu, 17 Dec 2015 23:28:47 GMT

Thanks, ballardw. The code posted above gives me desired result (it deletes from both data sets). I don't want to create temporary variale now (just for learning purpose) . I wanted to know how the data step worked. Since set statment in my example stacks a and b together sequentially, does "if " statement operate on each data sets sequentially. Basically, I want to see what happned during _n_=1?

Re: concatenating and subsetting data sets

ballardw — Thu, 17 Dec 2015 23:46:46 GMT

Test it by adding N=_n_; after the SET but before the if.

And temporary variables are just that, they would not be in the output, just like _n_ isn't unless assigned to another variable that isn't dropped from the data.