Merging Information From One Dataset To Another But Retaining Original Copies

[mkimmi]

Not sure if "merge" is the function I ultimately want, but it's what I tried....and flopped lol

So I have two data sets:

1. Base (goes on much longer than this)

A	B	C	ZIPCODE	YEAR	D
a	b	c	12345	2015	j
d	e	f	11011	2015	k
g	h	i	11102	2013	l
ch	cs	fm	11011	2015	kk

2. References: 

Z	Y	ZIPCODE	YEAR
m	p	12345	2015
n	q	11011	2015
o	r	11102	2013
s	t	90384	2013
u	v	10090	2016

Both of them have say two variables in common: Name is correct, format is the same, information is the same. (Zipcode and Year)

For each observation in Base dataset, I want to look up on the References table and find the matching Zipcode and Year. If and only if there is a match, then add the variables from References table to the Base dataset:

A	B	C	ZIPCODE	YEAR	D	Z	Y
a	b	c	12345	2015	j	m	p
d	e	f	11011	2015	k	n	q
g	h	i	11102	2013	l	o	r
ch	cs	fm	11011	2015	kk	o	r

However, there are

(1) repeated Zipcode and Year in Base dataset (11011, 2015) even though other variables have different information, so I want to be able to append the information while retaining that information in the References dataset for later observations that have the same zipcode/year.

and

(2) some Zipcode and Year in Reference dataset may not be present in Base dataset (e.g. 90384, 2013 and 10090, 2016), so I don't want these to merge into Base dataset with empty cells for other variables (since there are no observations in Base dataset with said zip code and year).

These are two issues I've been running into, and not sure how to solve...

Thanks in advance!!

[Shmuel]

You mentioned "Both of them have say two variables in common:". The only common variables I see are zipcode and year.

You mentioned "name is correct" - there is no variable named NAME.

In general you can use sql to join data from the two tables:

proc sql;
   create table want as
   select b.*   /* all BASE variables */
             ,r.<var1>
             .r.<var2>
            /* add as many REFERENCE variables as need */
   from BASE as b
   left join REFERENCE as r
   on (b.<common var-1> = r.<common var-1> and
         b.<common var-2> = r.<common var-2>
       /* compete any more conditions if need */
       )
      sort by b.<common var-1> , b.<common var-2>;
quit;

[mkimmi]

Quick clarification: "Both of them" -- both datasets of two common variables, whose names are correct -- as in ZIPCODE is ZIPCODE and YEAR is YEAR in both datasets

Thanks so much--will try this!! I'm finally getting a hang of proc sql and macros

[mkimmi]

I'm not sure why, but my BASE file is still the same when I proc print it, there are no new variables attached....

[Shmuel]

Please post the exact code you run, or even better the full log of the run showing the code and the messages.