topic Re: What's the best and more efficient method to map / define one variable based on three variables in SAS Enterprise Guide

What's the best and more efficient method to map / define one variable based on three variables

alepage — Fri, 21 Nov 2025 16:42:37 GMT

Hello,

Make it simple.

Let's say that I want to define one variable (ex: var4) based on three variables (ex: var1, var2, and var3), considering that based on the various combinaison of var1 var2 and var3, var4 could take many different values.

The first approach that I have in mind will be the if and condition ex:

if var1 eq something and var2 eq something else and var3 eq something then var4=specific_value

and so on.

There is also the proc format but I dont know how to define / map a new variable based on three variables.

Which approach is the more efficient

Re: What's the best and more efficient method to map / define one variable based on three variables

quickbluefish — Fri, 21 Nov 2025 17:22:17 GMT

I think you need to provide a better description (an example) of what you're trying to do.

Re: What's the best and more efficient method to map / define one variable based on three variables

Quentin — Fri, 21 Nov 2025 17:58:45 GMT

As you say, you can use PROC FORMAT to map one value to another.

Once I had a problem where I needed to map three values into another value. I started to build a new variable that would have the three values concatenated, and was going to build a format to map the concatenated value to my target value.

Then I remembered that SAS has hash tables. You can define a hash table that has 3 key values, and use that as a lookup.

As to whether a hash table is "best" or "more efficient", it really depends on your use case. There are lots of ways to do a lookup in SAS.

Re: What's the best and more efficient method to map / define one variable based on three variables

Ksharp — Sat, 22 Nov 2025 07:16:25 GMT

Check GRAYCODE() function:
https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/lefunctionsref/p1myrepjl958iin1ikhv6515cgp4.htm

Re: What's the best and more efficient method to map / define one variable based on three variables

Tom — Mon, 24 Nov 2025 03:53:23 GMT

If you want to do equality tests then a hash object is probably the simplest, as long as the number of possible values is small. Create a "truth" or "lookup" table with all of the possible values.

So if VAR1 to VAR3 could each have only two possible values then your LOOKUP table will have 8 observations. So something like:

data lookup;
  input var1 :$1. var2 var3 var4 :$8.;
cards;
A 1 1 1
A 1 1 2
A 2 1 3
A 2 2 4
B 1 1 5
B 1 2 6
B 2 1 7
B 2 2 8
;

Then you can load that into a HASH and use that to add VAR4 to your existing dataset (let's call it HAVE) and create a new dataset (let's call it WANT).

data want;
  set have lookup(keep=var4 obs=0);
  if _n_=1 then do;
    declare hash h(dataset:'lookup');
    h.definekey('var1','var2','var3');
    h.definedata('var4');
    h.definedone();
  end;
  if h.find() then call missing(var4);
run;

Re: What's the best and more efficient method to map / define one variable based on three variables

alepage — Tue, 25 Nov 2025 20:08:11 GMT

Can you please explain a litle bit more the hash approac and how does it works ?

Re: What's the best and more efficient method to map / define one variable based on three variables

Kurt_Bremser — Tue, 25 Nov 2025 22:01:40 GMT

I strongly encourage you to read Using the Hash Object to Store and Retrieve Data and come back with any further questions you have that need clarification.