topic Re: Joining tables in SAS Programming

Joining tables

anandas — Tue, 05 Sep 2017 14:12:37 GMT

Hello everyone

I have two tables. I need to join them and connect the columns. Below are the two source tables and the third table is the desired table.

In table 1 the first_term is the term that student began that program.

Table 1

Id	first_term	Program
610	199601	Libarts
610	201001	Medtran
610	201102	Photo

In table 2, the "term" column has all the terms the student has registered for a class. if the starting term in the "term" column is greater than or equal to the "first_term" in table 1, then it means that class (CRN number) is tied to that program. For example, the first program the student began is LIBARTS in 199601 (table 1). In table 2, we can see after 199601, this student took courses till the term reaches to the beginning of the new program which is 201001. This means the first 5 observations belong to the LIBARTS program.

Table 2

Id	Term	CRN
610	199602	211
610	199702	213
610	200501	102
610	200903	302
610	200903	321
610	201001	124
610	201001	153
610	201001	150
610	201001	142
610	201002	225
610	201002	242
610	201002	244
610	201003	316
610	201101	103
610	201101	121
610	201101	142
610	201101	145
610	201102	242
610	201102	242
610	201103	315
610	201201	140
610	201201	140
610	201201	150
610	201202	211
610	201202	241
610	201202	242

Basically, the joined table should look like below.

Id	Term	CRN	Program
610	199602	211	Libarts
610	199702	213	Libarts
610	200501	102	Libarts
610	200903	302	Libarts
610	200903	321	Libarts
610	201001	124	Medtran
610	201001	153	Medtran
610	201001	150	Medtran
610	201001	142	Medtran
610	201002	225	Medtran
610	201002	242	Medtran
610	201002	244	Medtran
610	201003	316	Medtran
610	201101	103	Medtran
610	201101	121	Medtran
610	201101	142	Medtran
610	201101	145	Medtran
610	201102	242	Photo
610	201102	242	Photo
610	201103	315	Photo
610	201201	140	Photo
610	201201	140	Photo
610	201201	150	Photo
610	201202	211	Photo
610	201202	241	Photo
610	201202	242	Photo

Any help will be greatly appreciated. Thanks.

Re: Joining tables

FredrikE — Tue, 05 Sep 2017 15:02:03 GMT

This might be one way to solve it 🙂

data one;

length id 8 first_term $6 program$32;

input id first_term program;

datalines;

610 199601 Libarts

610 201001 Medtran

610 201102 Photo

;

run;

data two;

length Id 8 Term $6 CRN $32;

input id term crn;

datalines;

610 199602 211

610 199702 213

610 200501 102

610 200903 302

610 200903 321

610 201001 124

610 201001 153

610 201001 150

610 201001 142

610 201002 225

610 201002 242

610 201002 244

610 201003 316

610 201101 103

610 201101 121

610 201101 142

610 201101 145

610 201102 242

610 201103 315

610 201201 140

610 201201 150

610 201202 211

610 201202 241

610 201202 242

;

run;

proc sort data=one;

by id descending first_term;

run;

data one;

set one;

length fromdate todate $6;

fromdate = first_term;

todate = lag(first_term);

if todate = '' then todate = '599901';

run;

proc sql;

create table tog as

select two.id, two.term, two.crn, one.program from one

inner join two

on one.id = two.id

and one.fromdate <= two.term < one.todate

;

run;

Re: Joining tables

anandas — Tue, 05 Sep 2017 15:33:02 GMT

Thank you for your response. This code works if there is only one student. But I have many students. The 'todate' has to be sorted by id? Currently the 'lag' function picks the date from previous observation which means it overlaps to another student? Hope i am making sense?

Re: Joining tables

FredrikE — Tue, 05 Sep 2017 16:03:08 GMT

Yes of, course, quick and wrong....:)

This is untested, but try:

data one;

set one;

by id descending first_term;

length fromdate todate $6;

fromdate = first_term;

todate = lag(first_term);

if first.id then todate = '599901';

run;

Re: Joining tables

anandas — Tue, 05 Sep 2017 16:06:54 GMT

Thanks Fredrik. Tried that, did not work as it still picks the terms from previous ids.

Re: Joining tables

FredrikE — Wed, 06 Sep 2017 06:28:06 GMT

Hmm, I think it works when I run it 🙂

Does your code look like this:

data one;

length id 8 first_term $6 program$32;

input id first_term program;

datalines;

610 199601 Libarts

610 201001 Medtran

610 201102 Photo

510 199701 Libarts1

510 201101 Medtran1

510 201302 Photo1

;

run;

data two;

length Id 8 Term $6 CRN $32;

input id term crn;

datalines;

610 199602 211

610 199702 213

610 200501 102

610 200903 302

610 200903 321

610 201001 124

610 201001 153

610 201001 150

610 201001 142

610 201002 225

610 201002 242

610 201002 244

610 201003 316

610 201101 103

610 201101 121

610 201101 142

610 201101 145

610 201102 242

610 201103 315

610 201201 140

610 201201 150

610 201202 211

610 201202 241

610 201202 242

510 199702 211

510 199802 213

510 200601 102

510 200904 302

510 200905 321

510 201101 124

510 201101 153

510 201101 150

510 201101 142

510 201102 225

510 201102 242

510 201102 244

510 201103 316

510 201401 103

510 201401 121

;

run;

proc sort data=one;

by id descending first_term;

run;

data one;

set one;

by id descending first_term;

length fromdate todate $6;

fromdate = first_term;

todate = lag(first_term);

if first.id then

todate = '599901';

run;

proc sql;

create table tog as

select two.id, two.term, two.crn, one.program from one

inner join two

on one.id = two.id

and one.fromdate <= two.term < one.todate

;

run;

Re: Joining tables

anandas — Wed, 06 Sep 2017 13:34:51 GMT

Thanks Fredrik. This code worked. I swear I did the same way yesterday and it did not work. I am guessing I may have something wrong.

I created my own code but seems more complicated. Below is my code;

data sgbstdn1a (drop=x);
set sgbstdn;
by sgbstdn_pidm;
length fromdate todate $6;
fromdate = first_term;
x=lag(first_term);
if first.sgbstdn_pidm then do;
todate='999999';
end; else do;
todate=x;
end;
run;

Re: Joining tables

FredrikE — Wed, 06 Sep 2017 13:42:12 GMT

The important thing is to sort in descending date order since first.term actually is the last one, and you need to go in reverse order to get the date intervals correct 🙂

//Fredrik

Re: Joining tables

anandas — Wed, 06 Sep 2017 13:45:45 GMT

Thank you! Makes sense!