topic Re: Date Merge in SAS Programming

Date Merge

rajd1 — Mon, 06 Jul 2020 22:48:42 GMT

Hi all,

I am trying to merge dates by the closest day. I just need help tweaking the code a little bit. I tried to look around but could not find a way to accomplish what i need. Any help is appreciated!

Here are the two datasets:

data aa;
input id order MMDDYY10.;
format order MMDDYY10.;
cards;
1 7/16/2016
1 10/2/2017
2 7/17/2016
3 7/18/2016
4 12/15/2018
4 1/19/2016
5 2/21/2017
5 5/19/2018
5 8/24/2018
6 10/2/2016
6 3/10/2018
7 5/9/2018
;

data bb;
input id received MMDDYY10.;
format received MMDDYY10.;
cards;
1 07/26/2016
1 10/22/2017
2 8/1/2016
3 8/3/2016
3 8/3/2016
4 1/6/2019
5 9/8/2018
6 3/25/2018
7 5/10/2018
7 10/10/2019
;

Data want

ID	Order	Received
1	7/16/2016	7/26/2016
1	10/2/2017	10/22/2017
2	7/17/2016	8/1/2016
3	7/18/2016	8/3/2016
4	12/15/2018	1/6/2019
4	1/19/2016
5	2/21/2017
5	5/19/2018
5	8/24/2018	9/8/2018
6	10/2/2016
6	3/10/2018	3/25/2018
7	5/9/2018	5/10/2018
7		10/10/2019

Here is my code:

proc sql;
create table want as
select distinct a.*,b.received
from aa a left join bb b
on a.id=b.id and order<=received
group by a.id,order;
quit;

I am missing the "having" keyword after "group by". Or would this be easier in data step?

Re: Date Merge

smantha — Mon, 06 Jul 2020 23:02:40 GMT

proc sql;
create table want as
select distinct a.*,b.received
from aa a left join bb b
on a.id=b.id and order<=received
and (received-order) <= <threshold_value_in_days>
group by a.id,order;
quit;

Re: Date Merge

rajd1 — Mon, 06 Jul 2020 23:27:25 GMT

This works but the missing order for ID#7 does not show in the output dataset.

Re: Date Merge

smantha — Mon, 06 Jul 2020 23:41:47 GMT

proc sql;
create table want as
select distinct a.*,b.received
from aa a left join bb b
on a.id=b.id and order<=received
and ( (received-order) <= <threshold_value_in_days> or missing(order))
group by a.id,order;
quit;

Re: Date Merge

smantha — Mon, 06 Jul 2020 23:44:19 GMT

proc sql;
create table want as
select distinct a.*,b.received
from aa a full join bb b
on a.id=b.id and order<=received
and ((received-order) <= <threshold_value_in_days> or missing((received-order)))
group by a.id,order;
quit;

Re: Date Merge

Kurt_Bremser — Tue, 07 Jul 2020 08:15:13 GMT

A slight change to your code, so that id is set correctly for the "received" that do not have a matching order:

proc sql;
create table want1 as
  select distinct
    coalesce(a.id,b.id) as id,
    a.order,
    b.received
  from aa a full join bb b
  on a.id = b.id and order <= received
    and ((received - order) <= 90 or missing((received - order)))
  group by calculated id, order
;
quit;

Re: Date Merge

rajd1 — Tue, 07 Jul 2020 15:35:14 GMT

This is great! Just as a thought if there was a way not to use the threshold value in days...

Re: Date Merge

Kurt_Bremser — Tue, 07 Jul 2020 16:48:18 GMT

@rajd1 wrote:
This is great! Just as a thought if there was a way not to use the threshold value in days...

You can use the INTCK function to computee the distance in time.