longitudinal cohort data manipulation

Justin_Lang — Tue, 11 Feb 2020 15:39:19 GMT

Hi SAS community,

I have a longitudinal cohort dataset. I need to keep only participants with urban = 1 at baseline.

Dataset1

UniqID	year	urban
101	2001	1
101	2002	1
101	2003	1
101	2004	1
102	2001	0
102	2002	0
102	2003	1
102	2004	1
103	2001	0
103	2002	0
103	2003	0
103	2004	1
104	2001	1
104	2002	0
104	2003	0
104	2004	0

Needs to look like this:

UniqID	year	urban
101	2001	1
101	2002	1
101	2003	1
101	2004	1
104	2001	1
104	2002	0
104	2003	0
104	2004	0

Can someone please help me!

Justin

Re: longitudinal cohort data manipulation

ballardw — Tue, 11 Feb 2020 15:47:32 GMT

@Justin_Lang wrote:

Hi SAS community,

I have a longitudinal cohort dataset. I need to keep only participants with urban = 1 at baseline.

Dataset1

UniqID year urban

101 2001 1

101 2002 1

101 2003 1

101 2004 1

102 2001 0

102 2002 0

102 2003 1

102 2004 1

103 2001 0

103 2002 0

103 2003 0

103 2004 1

104 2001 1

104 2002 0

104 2003 0

104 2004 0

Needs to look like this:

UniqID year urban

101 2001 1

101 2002 1

101 2003 1

101 2004 1

104 2001 1

104 2002 0

104 2003 0

104 2004 0

Can someone please help me!

Justin

See if this gets you started.

/* assumes input set sorted by ID and year*/
data want;
  set have;
  by uniqid ;
  retain flag;
  if first.uniqid then flag= (urban=1);
  if flag;
  drop flag;
run;

From the way you present your example data I suspect that your Urban value might actually be character. If the above code gives you messages about conversion to character messages then likely you should change the urban=1 to urban='1'.

The key part is the BY processing sets an automatic variable referenced as first.uniqid that is true for the first record with the value. The Flag = does a logical comparison of the Urban value an if true will have a value of 1(true) or 0 (false). The Retain statement keeps the value of flag for the next iteration of the data step. The If Flag subsets the data to have only the values where Flag is true.

Since you really wouldn't need the value of flag later I drop it.

You could test the logic of the code without the IF flag; and Drop Flag ; statements to see the result of the logic.

Re: longitudinal cohort data manipulation

Justin_Lang — Tue, 11 Feb 2020 16:02:20 GMT

Thanks for the reply. This seems to have worked for me!

topic longitudinal cohort data manipulation in SAS Enterprise Guide

longitudinal cohort data manipulation

Re: longitudinal cohort data manipulation

Re: longitudinal cohort data manipulation