topic Re: How to label rows that don't match in SAS Programming

How to label rows that don't match

bourdeax — Wed, 29 Apr 2020 13:48:34 GMT

Hello Everyone

First, please take a look at this data set:

ID College1 College2 NewVariable
1 A A YES
1 A B YES
1 A A YES
2 A B NO
2 A B NO
2 A B NO
3 A A YES
3 A A YES
3 A A YES

My goal is to create NewVariable. Basically, if none of the values of College2 = College1, then NewVariable = No. If all the values of College2 = College1, or if some of the values of College2 = College1, then NewVariable = Yes.

Thank you!

Re: How to label rows that don't match

PaigeMiller — Wed, 29 Apr 2020 14:12:27 GMT

proc sql;
	create table want as select *,sum(college1=college2) as num
		from have group by id;
quit;

basically, this creates a new variable NUM which counts the number of matches. I will leave it up to you to turn the 0 match case into "NO" and the cases where there is 1 or more matches into "YES".

Re: How to label rows that don't match

Kurt_Bremser — Wed, 29 Apr 2020 14:23:17 GMT

Data step solution:

data want;
set have;
by id;
retain newvar;
if first.id then newvar = 'N';
if college1 = college2 then newvar = 'Y';
if last.id;
keep id newvar;
run;

If needed, you can join that back to the original dataset.

Re: How to label rows that don't match

bourdeax — Wed, 29 Apr 2020 14:47:57 GMT

Thank you, Paige. That worked. Also, I appreciate how you answered my main question but left some work for me to do. That is a good teaching technique.

Re: How to label rows that don't match

bourdeax — Wed, 29 Apr 2020 14:49:54 GMT

Also, in case anyone wants to know what I did after Paige's section of code:

data want;
set have;
if num = 0 then flag = "N";
else flag = "Y";
run;