topic Re: How to select first obs of multiple records in SAS Programming

How to select first obs of multiple records

Kiko — Thu, 03 Aug 2017 15:03:00 GMT

Hello

I have a variable called Plural in my dataset (numeric; 1-8; 9=unknown). Since for those w. Plural >1 will have duplicate values for the variables I am interested in, I want SAS to keep only one record (doesn't matter which) when Plural >1. Does anyone know how to do this?

Thank you!

Re: How to select first obs of multiple records

Shmuel — Thu, 03 Aug 2017 15:08:32 GMT

You haven't post test data.

You can use proc sort with option NODUPKEY to achieve one observation of

any plural occurence.

Re: How to select first obs of multiple records

Kiko — Thu, 03 Aug 2017 15:16:36 GMT

Thank you for your reply. But if I PROC SORT w. NODUPKEY by Plural SAS will get rid of the most obs because most of the obs in my dataset has Plural=1. I want SAS to pick one record only when Plural >1.

Re: How to select first obs of multiple records

Shmuel — Thu, 03 Aug 2017 15:24:11 GMT

please post an example - some test data to explain what you have and the desired result.

Re: How to select first obs of multiple records

Kiko — Thu, 03 Aug 2017 16:00:37 GMT

ID DOB DATE1 DURATION PLURAL FEEDING

1 2/2/85 4/8/16 30 2 YES

2 2/2/85 4/8/16 30 2 NO

3 4/2/79 5/1/14 1 3

4 4/2/79 5/1/14 1 3 NO

5 4/2/79 5/1/14 1 3

6 8/1/12 5/1/16 10 1 NO

7 1/2/14 10/1/17 2 1

8 8/1/12 8/1/15 1 1 NO

So here's what my dataset kind of looks like.

ID and Feeding are collected from child. DOB, DATE1, DURATION, PLURAL are from mother. I am interested in the variable DURATION. Since DURATION is assigned to mother I'd like to keep only one record of this variable when Plural is greater than 1. If Plural=1 then there is only one record so I don't have to worry about it.

Hope this makes sense! Thank you!

Re: How to select first obs of multiple records

Shmuel — Thu, 03 Aug 2017 16:15:18 GMT

Check next code:

proc sort data=have out=want nodupkey;
  by DOB DATE1 PLURAL;
run;

Do you prefer the record with FEEDIN=YES or FEEDING=NO ?

If you have a preference then use next code:

proc sort data=have out=temp;
  by DOB DATE1 PLURAL 
       {ascending or descending } FEEDING;
run;

data want;
 set temp;
  by DOB DATE1 PLURAL;
       if first.plural;
run;