Merging the two datasets - sas

Question

Dataset A:
Company_Name Match Sales EMPS
1234 0 0 0
1234 0 0 0
1234 0 0 0
5678 0 0 0
5678 0 0 0
5678 0 0 0
9123 9123 500 2
9123 9123 500 2
9123 9123 500 2
Dataset B:
Company_Name Match Sales EMPS
1234 1234 600 10
1234 1234 600 10
1234 1234 600 10
5678 5678 900 56
5678 5678 900 56
5678 5678 900 56
I am trying to merge the above tables using proc sql, and here is the desired output
Dataset A:
Company_Name Match Sales EMPS
1234 1234 600 10
1234 1234 600 10
1234 1234 600 10
5678 5678 900 56
5678 5678 900 56
5678 5678 900 56
9123 9123 500 2
9123 9123 500 2
9123 9123 500 2
However, when I try to do a join, it only takes the first table's values. I know I should do a case statement somewhere, but not sure how. For example, since datasetb had values for company_name=1234, the final output should capture that, and if there are no values, it should take the column values of the first table, if that makes sense
proc sql;
create table merge_table as
select a.*,b* from dataseta as a inner join datasetb as b on (a.company_name=b.company_name);quit;

Answer 1

Use the COALESCE() function to code your preference for B values over A value.
create table merge_table as
select a.company_name
, coalesce(b.match,a.match) as match
, coalesce(b.sales,a.sales) as sales
, coalesce(b.EMPS,a.EMPS) as EMPS
from dataseta as a
inner join datasetb as b
on (a.company_name=b.company_name)
;
But your example has repeats for COMPANY_NAME in both datasets. How do you want to handle that? Currently it will match each of the three records from A for company 1234 with each of the three records from B for company 1234 and produce 9 records for that company in the result set. You need some other variable(s) to include in the join condition so that a performs a 1 to 1 match (or at least a 1 to N match) instead of the current N to M match.

Answer 2

Assuming same number of zero rows and non-zero rows to replace, consider a union query to stack non-zeros and other dataset:
proc sql;
create table merge_table as
select b.Company_Name, b.Match, b.Sales, b.EMPS
from datasetb as b
union
select a.Company_Name, a.Match, a.Sales, a.EMPS
from dataseta as a
where (a.Match + a.Sales + a.EMPS) ^= 0;
quit;