How to Calculate Percentage with a PROC SQL

[turcay]

Hello everyone, I think I have a very simple question but I can't handle it. I'm trying to get percentage of values of the variables. I think I'm so close the desired output but I want to be sure. Could you help me, please ?

Sample Data ;

DATA Have;
Length Variable_ID $ 3 Variable $ 3;
Infile Datalines Missover Dlm=",";
Input Variable_ID Variable;
Datalines;
001,TUR
001,TUR
002,TUR
002,TUR
003,ENG
003,ENG
004,GER
004,GER
005,FRA
005,FRA
;
Run;
PROC SQL;
Create Table Want As
Select
Variable,
Count(1) as CountOfVariable
/*Calculated CVariable/100 as Percent format=percent8.2*/
From Have
group by Variable;
QUIT;

Desired Output;

[Jagadishkatam]

proc sql;
select count(*) into: obs from have; 
create table test as select variable, (count(variable)/&obs) as var format=percent8.2, count(*) as count from have group by variable;
quit;

[PaigeMiller]

I feel the solutions provided by @Jagadishkatam and also by @mohamed_zaki could give the wrong result in some situations.

If there are missing values in variable, then the variable &obs will count the missings, and then in the next line, (count(variable)/&obs) will have the wrong denominator.

[mohamed_zaki]

PROC SQL noprint;
select nobs into :nobs from dictionary.tables 
where libname="WORK" and memname="HAVE";

Create Table Want As
Select
Variable, Count(1)/&nobs as Percent format=percent8.2
From Have
group by Variable;
QUIT;

[Ksharp]

DATA Have;
Length Variable_ID $ 3 Variable $ 3;
Infile Datalines Missover Dlm=",";
Input Variable_ID Variable;
Datalines;
001,TUR
001,TUR
002,TUR
002,TUR
003,ENG
003,ENG
004,GER
004,GER
005,FRA
005,FRA
;
Run;
PROC SQL;
Create Table Want As
Select
Variable,
count(*)/(select count(*) from have) as Percent format=percent8.2
From Have
group by Variable;
QUIT;

[turcay]

Hello,

Thank you all of them,I missed out joining with another table while i did this process according to the following conditions.

Conditions

Have1.Number=1
Have2.Number=1

DATA Have;
Length Variable_ID $ 3 Variable $ 3 Number 8;
Infile Datalines Missover Dlm=",";
Input Variable_ID Variable Number;
Datalines;
001,TUR,1
001,TUR,1
002,TUR,0
002,TUR,0
003,ENG,1
003,ENG,1
004,GER,1
004,GER,1
005,FRA,1
005,FRA,1
;
Run;
DATA Have2;
Length Variable $ 3 Number 8;
Infile Datalines Missover Dlm=",";
Input Variable Number;
Datalines;
TUR,1
TUR,1
TUR,0
TUR,0
ENG,1
ENG,1
GER,1
GER,1
FRA,1
FRA,1
;
Run;
PROC SQL;
CREATE TABLE Want AS
SELECT 
H1.Variable,
COUNT(*)/(SELECT count(*) FROM Have) AS Percent FORMAT=Percent8.2
FROM Have H1 
INNER JOIN Have2 H2
ON H1.Variable=H2.Variable
WHERE H2.Number=1
AND H1.Number=1
ORDER BY H1.Variable;
QUIT;

Now my desired output as below, Do you have an idea ?

Thanks.

[Ksharp · Accepted solution]

I am not quit sure what you are trying to do .

DATA Have;
Length Variable_ID $ 3 Variable $ 3 Number 8;
Infile Datalines Missover Dlm=",";
Input Variable_ID Variable Number;
Datalines;
001,TUR,1
001,TUR,1
002,TUR,0
002,TUR,0
003,ENG,1
003,ENG,1
004,GER,1
004,GER,1
005,FRA,1
005,FRA,1
;
Run;
DATA Have2;
Length Variable $ 3 Number 8;
Infile Datalines Missover Dlm=",";
Input Variable Number;
Datalines;
TUR,1
TUR,1
TUR,0
TUR,0
ENG,1
ENG,1
GER,1
GER,1
FRA,1
FRA,1
;
Run;
PROC SQL;
CREATE TABLE Want AS
SELECT 
H1.Variable,
COUNT(*)/(SELECT count(*)  FROM Have H1 
INNER JOIN Have2 H2
ON H1.Variable=H2.Variable
WHERE H2.Number=1 AND H1.Number=1 ) AS Percent FORMAT=Percent8.2
FROM Have H1 
INNER JOIN Have2 H2
ON H1.Variable=H2.Variable
WHERE H2.Number=1
AND H1.Number=1
group BY H1.Variable;
QUIT;

[RW9]

How about:

data have;
  length variable_id $ 3 variable $ 3 number 8;
  infile datalines missover dlm=",";
  input variable_id variable number;
datalines;
001,TUR,1
001,TUR,1
002,TUR,0
002,TUR,0
003,ENG,1
003,ENG,1
004,GER,1
004,GER,1
005,FRA,1
005,FRA,1
;
run;
data have2;
  length variable $ 3 number 8;
  infile datalines missover dlm=",";
  input variable number;
datalines;
TUR,1
TUR,1
TUR,0
TUR,0
ENG,1
ENG,1
GER,1
GER,1
FRA,1
FRA,1
;
run;

data inter;
  set have (drop=variable_id) have2;
  if number;
run;

proc sql;
  create table WANT as
  select  distinct
          A.VARIABLE,
          (select count(VARIABLE) from INTER where VARIABLE=A.VARIABLE) / (select count(VARIABLE) from INTER) as PERCENT format=percent8.2
  from    INTER A;
quit;

[LinusH]

It's fun to see how to solve things in different ways.

But, as long as we are in the SAS world, remember to use the best tool for the situation, and for this situation PROC FREQ is better suited (IMHO) than SQL.

[RW9]

Yes, but how often is it used...

/humour

[turcay]

Thank you all for helping me 🙂

[amandaokello]

I like your "sass"!

[Harmandeep]

This problem can be simplified by using a subquery for the count of total number of variable as :

DATA Have;
Length Variable_ID $ 3 Variable $ 3;
Infile Datalines Missover Dlm=",";
Input Variable_ID Variable;
Datalines;
001,TUR
001,TUR
002,TUR
002,TUR
003,ENG
003,ENG
004,GER
004,GER
005,FRA
005,FRA
;
Run;

Proc sql;

create table want as
select Variable, (Count(Variable) / (Select Count(Variable) from have)) * 100 as Percent
from have
group by Variable;

quit;