DATA Step, Macro, Functions and more

Create ranks using SQL

Accepted Solution Solved
Reply
Occasional Contributor
Posts: 12
Accepted Solution

Create ranks using SQL

[ Edited ]

Question : I have a dataset with department names, employee names, salary. I have to find out the 5th largest salary of employee along with its department by using Proc SQL (I know the procedure by using Data step but I want to know by using Proc SQL). Please help.

 

Dept_Name Emp_name Salary
Economics Naveen 1070
statistics Praveen 1157
Biology Manju 1286
marketing Chethan 1396
Economics vinay 1110
statistics sanjay 1318
Biology harshi 1376
marketing manu V 1179
Biology manu S 1002
statistics manu P 1442

Accepted Solutions
Solution
‎10-09-2017 06:58 AM
Trusted Advisor
Posts: 1,270

Re: Create ranks using SQL

Hi,

 

This is another way:

 

proc sql outobs=5;
create table want as
select * from have
order by salary desc;
select * from want
having salary=min(salary);
quit;

View solution in original post


All Replies
Respected Advisor
Posts: 2,660

Re: proc sql

Is this a homework problem?

 

I would advise to do this in a data step or PROC RANK. It's much simpler.

--
Paige Miller
Occasional Contributor
Posts: 12

Re: proc sql

Posted in reply to PaigeMiller

I want to know in Proc SQl.

I know the procedure in Data step by using rank and (_n_).

Super Contributor
Posts: 319

Re: proc sql

Post yout trial here and you will be helped for sure. Have you tried solve it by yourself?

Occasional Contributor
Posts: 12

Re: proc sql

Iam not that much good in Proc SQL. When I searched online I could't get exact solution for 5th largest. I got solution for 2nd largest, 3rd largest. So please help me how to get 5th Largest. Thanks in advance.

Respected Advisor
Posts: 2,660

Re: proc sql

Can you provide a link to the solution for 3rd largest? Why wouldn't that generalize to the 5th largest? Show us your code!

 

And while it is admirable to try to increase your SQL skills, this really isn't something that you should attempt to do in SQL, since SQL is simply a very poor choice in SAS to do this calculation.

--
Paige Miller
Occasional Contributor
Posts: 12

Re: proc sql

Posted in reply to PaigeMiller

The above problem is interviewed question. So Iam seeking help in this portal. Thanks.

Respected Advisor
Posts: 2,660

Re: proc sql

[ Edited ]

Naveen1 wrote:

The above problem is interviewed question. So Iam seeking help in this portal. Thanks.


The answer to the interview question is:

 

"It would be easier to do this in a DATA step or using PROC RANK. Doing this in PROC SQL is a very poor choice."

--
Paige Miller
Super Contributor
Posts: 320

Re: proc sql

1/ Sort by Salary (with a SELECT ... ORDER BY if you have to use SQL only)

2/ Use the monotonic function to get the fifth record.

Occasional Contributor
Posts: 12

Re: proc sql

Can I get elabprated syntax, as I informed earlier I dont have much knowledge on  Proc SQL.

Super User
Posts: 5,829

Re: proc sql

You have still not motivated why you want/need to do it in SQL.

If it's for learning purposes, don't do it, since given the answer already given, there are better ways to do this. Don't learn a less efficient way.

If you want to learn SQL, there must be tons of other exercises to go about.

Data never sleeps
Super User
Posts: 10,621

Re: Create ranks using SQL

Why you want to do this. it is really clumsy for SQL to get the largest n obs.
The following return the 5th largest weight for each sex.



proc sql;
select sex,name,weight,(select count(*) as n  from (select distinct sex,name,weight from sashelp.class) as b
          where b.sex=a.sex and b.weight<=a.weight) as n
 from sashelp.class as a
   where calculated n=5;
quit;

Solution
‎10-09-2017 06:58 AM
Trusted Advisor
Posts: 1,270

Re: Create ranks using SQL

Hi,

 

This is another way:

 

proc sql outobs=5;
create table want as
select * from have
order by salary desc;
select * from want
having salary=min(salary);
quit;

☑ This topic is solved.

Need further help from the community? Please ask a new question.

Discussion stats
  • 12 replies
  • 306 views
  • 2 likes
  • 7 in conversation