Random sorting when combining data

[Q1983]

data have;

length Teammate $50;

input Process $  Teammate $;

datalines

AAAA     John

AAAA    Charles

BBBB  Samuel

run;

input_record

LN     PROCESS      TEAMMATE

1        AAAA    

2        AAAA

3      BBBB

PROC SQL;

CREATE TABLE COMBINE AS

SELECT *

FROM HAVE  A

LEFT JOIN INPUT_RECORD  B

ON A.PROCESS = B.PROCESS

;QUIT

If I run this code process AAAA will be assigned to teammate    John for both 1 and 2.  Is there a way to assigned this randomly (John would get 1 and Charles would get 2).  The reason for this is because the customer provided a spreadsheet similar to input_record in which multiple people can be assigned the same process.

[kiranv_]

use a merge instead of join 

data have;

length Teammate $50;

input Process $  Teammate $;

datalines;

AAAA     John

AAAA    Charles

BBBB  Samuel
;
run;



data
input_record;
input 
LN     PROCESS  $    ;
datalines;
1      AAAA    
2      AAAA
3      BBBB
;
data want;
merge have(in=a) input_record(in=b);
by process;
if a;
run;

that should solve this

[mkeintz]

@kiranv_

Given the OP's description of the objective, this solution depends ENTIRELY on

1. Each dataset having the same order of by-groups.  (But that can be solved by a proc sort).
2. The two datasets having exactly then same number of records per by-group.  Otherwise, whichever merged data set has a smaller number of within-by-group records will have its final values "retained" through all the "extra" records from the other data set.
3. Each dataset having the same order within by-groups.  Otherwise, there is no apparent variable to validate a match.

In other words, can the OP trust his data providers to be that consistent?   Having a blank teammate variable, which would ordinarily be a match-key, diminishes one's confidence in the data provider.

[kiranv_]

you are so right @mkeintz

[r_behata]

A SQL Query would produce a Cartesian product  ( 5 rows in the sample output, 2 rows each for the repeated values) , are you sure is that is what you intend ?  .

If "John would get 1 and Charles would get 2" is the only criteria use a data step merge instead. Not sure what you meant by "Random" in the example.

[Q1983]

data have1;

length Process Teammate $50;

input Process $ Teammate$;

datalines;

POC James

POC Sam

POC Jane

run;

data have2;

length Process Teammate $50 ;

input ln Process $ Teammate $ ;

datalines;

12 POC James

55 POC

60 POC

;run;

data merge1;

merge have1 have2;

if process = 'POC';

run;

OUTPUT

ln Process Teammate

12 POC James

55 POC

60 POC

How can I modify the code to randomly assign the teammate in a sequencr of

James, Sam Jane

ln Process Teammate

12 POC James

55 POC Sam

60 POC Jane

72 POC James

88 POC Sam

99 POC Jane

....etc