Merging two datasets

[withpeople86]

I want to make a table by combing the following two tables. 

dy	dm	age_group	counts
1997	1	2	20
1997	1	3	30
1997	1	4	20
1997	2	2	5
1997	2	3	14
1997	2	4	23
1997	3	1	23
1997	3	2	22
1997	3	3	11
1997	3	4	5
1997	4	1	52
1997	4	2	34
1997	4	3	25
1997	4	4	23
1998	1	1	53
1998	1	2	5
1998	1	3	32
1998	1	4	20
1998	2	1	44
1998	2	2	42
1998	2	3	11
1998	2	4	5
1998	3	1	34
1998	3	2	20
1998	3	3	30
1998	3	4	32
1998	4	1	10
1998	4	2	23
1998	4	3	10
1998	4	4	20

table a

year	age_group	total	std
1997	1	3500	28
1997	2	6500	25
1997	3	7600	17
1997	4	24300	30
1998	1	3600	28
1998	2	6500	25
1998	3	7800	17
1998	4	24300	30

table B

What I want looks like this. 

ddate	age_group	counts	total	std
1997/01	1	0	3500	28
1997/01	2	20	6500	25
1997/01	3	30	7600	17
1997/01	4	20	24300	30
1997/02	1	0	3500	28
1997/02	2	5	6500	25
1997/02	3	14	7600	17
1997/02	4	23	24300	30
1997/03	1	23	3500	28
1997/03	2	22	6500	25
1997/03	3	11	7600	17
1997/03	4	5	24300	30
1997/04	1	52	3500	28
1997/04	2	34	6500	25
1997/04	3	25	7600	17
1997/04	4	23	24300	30
1998/01	1	53	3600	28
1998/01	2	5	6500	25
1998/01	3	32	7800	17
1998/01	4	20	24300	30
1998/02	1	44	3600	28
1998/02	2	42	6500	25
1998/02	3	11	7800	17
1998/02	4	5	24300	30
1998/03	1	34	3600	28
1998/03	2	20	6500	25
1998/03	3	30	7800	17
1998/03	4	32	24300	30
1998/04	1	10	3600	28
1998/04	2	23	6500	25
1998/04	3	10	7800	17
1998/04	4	20	24300	30

should I change the year and month variable to the calendar format?

How can I join these tables with proc sql?

Thanks for taking the time to look at it in advance.

[ChrisNZ]

Post your data as a data step.

This will allow us to know for example if you have SAS dates or strings.

[heffo]

I couldn't really get the missing values in there. It shouldn't be any problems but I guess I'm tired. So, I added a step in the middle to add the missing age_groups. 

data TableA;
input dy	dm 	age_group	counts;
datalines;
1997 1 2 20
1997 1 3 30
1997 1 4 20
1997 2 2 5
1997 2 3 14
1997 2 4 23
1997 3 1 23
1997 3 2 22
1997 3 3 11
1997 3 4 5
1997 4 1 52
1997 4 2 34
1997 4 3 25
1997 4 4 23
1998 1 1 53
1998 1 2 5
1998 1 3 32
1998 1 4 20
1998 2 1 44
1998 2 2 42
1998 2 3 11
1998 2 4 5
1998 3 1 34
1998 3 2 20
1998 3 3 30
1998 3 4 32
1998 4 1 10
1998 4 2 23
1998 4 3 10
1998 4 4 20
;;;
run;
 
data tableB;
input year	age_group	total	std;
datalines;
1997 1 3500 28
1997 2 6500 25
1997 3 7600 17
1997 4 24300 30
1998 1 3600 28
1998 2 6500 25
1998 3 7800 17
1998 4 24300 30
;;;
run;
data tableb;
	set tableb;
	do _i = 1 to 4;
		dm =_i;
		output;
	end;
	drop _i;
run;

proc sort data=tablea (rename=(dy=year));
	by year age_group dm;
run;
proc sort data=tableb;
	by year age_group dm;
run;
data want;
	retain ddate	age_group	counts	total	std;
	length ddate $ 10;
	update tableb tablea  ;
	by year age_group dm;
	ddate =cat(year,"/",dm);
	if counts = . then counts = 0;
	drop year dm;
run;
proc sort data=want;
	by ddate age_group;
run;

Stupid but it works? 🙂

[PGStats]

Do something like:

proc sql;
create table C as
select 
    mdy(dm,1,dy) as ddate format=yymms.,
    a.age_group,
    counts,
    total,
    std
from A inner join B on dy=year and A.age_group=B.age_group
order by ddate, age_group;
quit;