topic Re: Ranking character variable values in SAS Programming

Ranking character variable values

frupaul — Wed, 18 Jul 2018 11:42:14 GMT

Hi Everyone,

I have the following data set:

The above data set can be created with the code:

data Cust;
input Cust_name $ Manager1 $ Manager1 $ Manager3 $ Manager4 $ Manager5 $;
cards;
Jason Paul . . James Bond
Maxmil Lucien Peter . . Pan
Pogba . . Kit ARYNA Inieata
;

My objective is to create two variables:

- Highest_Ranked_Manager: the highest ranked manager of each customer

-Next_ranked_manager: The next highest ranked manager.

I believe this is self explanatory but the expected outcome table is:

Any ideas on how I can do this please?

Thanks

Re: Ranking character variable values

andreas_lds — Wed, 18 Jul 2018 13:02:41 GMT

If all manager-names have just one word, you could use:

data want;
   set work.CUST;

   length 
      ManagerList $ 50
      Highest_ranked_manager Next_Ranked_Manager $ 8
   ;

   ManagerList = catx(' ', of Manager:);

   Highest_ranked_manager = scan(ManagerList, countw(ManagerList));
   Next_Ranked_Manager = scan(ManagerList, countw(ManagerList)-1);

   keep Cust_name Highest_ranked_manager Next_Ranked_Manager;
run;

You need to adjust length of variables depending on the real cust dataset.

EDIT: If the names are more complex, than please add those cases to the example dataset and we shall see, if the code can be updated or arrays and loops are necessary.

Re: Ranking character variable values

Ksharp — Wed, 18 Jul 2018 13:08:45 GMT

data Cust;
input Cust_name $ Manager1 $ Manager1 $ Manager3 $ Manager4 $ Manager5 $;
cards;
Jason Paul . . James Bond
Maxmil Lucien Peter . . Pan
Pogba . . Kit ARYNA Inieata
;
data want;
 set cust;
 array x{*} $ manager:;
 do i=dim(x) to 1 by -1;
  if not missing(x{i}) then do;highest=x{i};call missing(x{i});leave;end;
 end;
 do i=dim(x) to 1 by -1;
  if not missing(x{i}) then do;next_highest=x{i};leave;end;
 end;
 drop i manager:;
 run;
 
 proc print noobs;run;

Re: Ranking character variable values

Tom — Wed, 18 Jul 2018 15:03:08 GMT

Good idea. You can simplify and make it more robust.

data want;
   set work.CUST;
   length Highest_ranked_manager Next_Ranked_Manager $50 ;
   Highest_ranked_manager = scan(catx('|', of Manager:), -1,'|');
   Next_Ranked_Manager = scan(catx('|', of Manager:), -2,'|');
run;

If you don't know what length to define the new variables you could let SAS figure it out be replacing the LENGTH statement with two assignment statements. That will make the length match the length of the first variable.

   Highest_ranked_manager = manager1;
   Next_Ranked_Manager = manager1;

Re: Ranking character variable values

frupaul — Wed, 18 Jul 2018 22:22:13 GMT

HI Andreas, Thanks for the proposed solution. It worked but i got a warning. Im guessing it was because of the use of the countw function. However, when i used the scan(ManagerList, -1) and scan(ManagerList, -2) it worked without a warning