How do I exclude a record from being considered for matching if already matched

[constliv]

@Patrick

Thank you very much indeed for your detailed description of the logic. I am very grateful.

[Patrick]

@constliv

And here the link to the documentation (ignore the Java section):

https://support.sas.com/documentation/cdl/en/lecompobjref/69740/HTML/default/viewer.htm#titlepage.ht...

Also: I've used in the hash declare statement the dataset:'<table name>' directive.

dcl hash h1(dataset:'a(rename=(id=id_a balance=balance_a))', multidata:'y');

Most examples in the documentation don't use this directive. What it does: It instructs SAS not only to define the hash structure (the key and data variables) but also to directly load data into the hash from the table as defined via dataset:'<table name>'.

Most examples won't do this but will only define the hash table structure and then use the add() method to actually load data into the hash. Don't let this confuse you. These are just different ways how you can populate a hash table and both methods have their use cases. You'll get it once you start using hashes. It takes a bit but it's very well worth learning as it opens possibilities you don't have without hashes.

One common use case: You have a big unsorted base table and you need to look up variables over a composite key in a small table. Using a hash table removes the need to sort the big table which can make a big difference in regards of performance. 

A standard case is having a huge fact table looking up values in multiple dimension tables over different keys. If you load the dimension tables into SAS hashes then you can implement this look-ups in a single data step without having to sort the fact table multiple times (whether sorting is done explicitely or implicitely via a SQL join).