Splitting a dataset by a condition

[mmaximos]

Hello all, 

I'm looking to split a datasets based on one of the variables. Currently in the drug column I have two medications: drug a and drug b, I want to have one dataset include just drug a and the other just drug b.

The code I tried is as follows but no success. Best way to correct this?

Thank you

Data mydata;
	Set sample;
	Where location= "country" and code= "anti-infective";
	Keep drug vol time_period;
Run;

data mydata.a;
set mydata;
keep drug= a;
run;

data mydata.b;
set mydata;
keep drug= b;
run;

[PaigeMiller]

Generally, splitting up a data set is not necessary and not productive. So let's abandon that idea.

If you want to perform an analysis using a certain PROC on part of the data set, you can do this:

proc something data=mydata(where=(drug='a'));

or better yet, you can use BY statements to have analyses performed on both drug A and drug B with one piece of code.

proc something data=mydata;
     by drug;
     /* Other statements for the PROC go here */
run;

which assumes that MYDATA is sorted by DRUG

[mmaximos]

Amazing, thank you! I'll give this a try. 

[Reeza]

To answer your question directly as asked, KEEP selects variables in a data set not observations.
WHERE filters observations in a data sets. Changing KEEP to WHERE in your code will get you the desired results. 

data mydata.a;
set mydata;
where drug= "a"; *note that this is case sensitive as well;
run;

data mydata.b;
set mydata;
where drug= "b";
run;

However, @PaigeMiller is correct, you usually do not want to do this, his solution is preferable. 

[mmaximos]

Thank you

[Kurt_Bremser]

The KEEP statement and the KEEP= dataset option control which variables go into the output, not which observations. Use a WHERE statement, like you did in your first step.

[mmaximos]

Thank you!