topic Re: Rank using multiple by variables in SAS Programming

Rank using multiple by variables

UdayGuntupalli — Mon, 02 Jul 2018 22:25:41 GMT

All,
I am looking for some help in trying to rank a data set using multiple by-variables. I am attaching a sample script that I tried to put together, however, I am not sure where I am going wrong. I am trying to follow the example provided here (http://support.sas.com/documentation/cdl/en/proc/65145/HTML/default/viewer.htm#p12aek9f6xhl1zn1puuuot19wqi7.htm) . Obviously, I am doing something wrong, however I am not familiar with Proc Rank or Proc Sort well enough to understand what I am doing wrong. If Var1 is my first grouping variable, Var2 is my second grouping variable and TestVal is the value I want to compute ranks of my dataset, can some body please guide me and explain what I am doing wrong.

data RankQuestion;
   input Var1 1-2 Var2 3-4 TestVal 5-9;
   datalines;
	1 1 0.42
	1 2 0.62
	1 3 0.22
	5 4 0.12
	5 6 0.11
	5 9 0.13
	8 7 0.99
	8 10 0.84
	8 11 0.85
	;
	
proc sort data = RankQuestion out = RawSort; 
	by Var1 Var2 TestVal; 
run; 

proc rank data = RawSort out = RankRes ties = low descending;
	by Var1 Var2; 
	var TestVal; 
	ranks newrank; 
run;

proc print data =  RankRes ; 
	by Var1 Var2 newrank; 
	title 'Testing Rank By Groups';
run;

Re: Rank using multiple by variables

Astounding — Mon, 02 Jul 2018 23:06:20 GMT

You'll have to show what you want the result to be. It's too much of a leap of faith to show us code that doesn't work right, and hope we can figure out what "work right" means.

Re: Rank using multiple by variables

UdayGuntupalli — Mon, 02 Jul 2018 23:50:05 GMT

@Astounding,
Var1 Var2 TestVal Rank

1 3 0.22 1
1 1 0.42 2

1 2 0.62 3

5 6 0.11 1
5 4 0.12 2

5 9 0.13 3

Here is an example of the desired output.

Re: Rank using multiple by variables

novinosrin — Tue, 03 Jul 2018 00:16:02 GMT

data RankQuestion;
   input Var1  Var2  TestVal ;
   datalines;
	1 1 0.42
	1 2 0.62
	1 3 0.22
	5 4 0.12
	5 6 0.11
	5 9 0.13
	8 7 0.99
	8 10 0.84
	8 11 0.85
	;
	
proc sort data = RankQuestion out = RawSort; 
	by Var1  testval; 
run; 
proc rank data = RawSort out = RankRes   ties=low; ;
	by Var1 ; 
	var   TestVal ; 
	ranks newrank1 ; 
run;

Re: Rank using multiple by variables

PGStats — Tue, 03 Jul 2018 00:17:35 GMT

You seem to want to order by TestVal within Var1 groups. What should be the role of Var2?