How to delete top and bottom value within a group

[Crubal]

Hi,

I have a dataset called data_1, I would like to delete smallest on largest 'Room_Rate' from the data within each (Brand, HTL_CD, Prod_Class_Name, RM_NBR, RM_TYP, REF_ROOM_TYP, SEASON_ID, DY_TYP) combination. 

For the example as below (data_1), I would like to delete the first and last row.

 

If I used the following syntax: 

Proc Sql;
		Create Table data_2 as select 
			 Brand 
			,Htl_Cd 
			,Prod_Class_Name 
			,RM_TYP 
			,REF_RM_TYP 
			,Season_Id 
			,DY_TYP 
			,Room_Rate 
		from data_1

		group by 1, 2, 3, 4, 5, 6, 7
		having Room_Rate > min(Room_Rate) and Room_Rate < max(Room_Rate)
		order by 1, 2, 3, 4, 5, 6, 7;
	Quit;

First two rows are deleted as they both have 'Room_Rate' $93. as well as the last row with 'Room_Rate' $144. 

But I do would like to keep one of the $93 row. 

Is there any way to handle it?

Thanks! 

[PaigeMiller]

If you have sorted by room_rate, then just delete the first and last observation

data want;
    set have end=eof;
    if _n_>1 and not eof then output;
run;

[Crubal]

Thanks @PaigeMiller

In this way, first observation and last observation could be removed. 

However, it does not apply to each group, (other Season_Id, DY_TYP, ...)

[PaigeMiller]

You have NOT shown us data with groups. We can't read your mind. We can't provide a solution for data you didn't give us.

Show us data with groups ... or simply change the solution I gave you to include FIRST. and LAST. logic, and the solution I gave works fine.

[Crubal]

Hi @PaigeMiller

I modified it to :

	Data data_2;
		Set data_1; 
	     by  Brand 
			Htl_Cd 
			Prod_Class_Name 
			RM_TYP 
			REF_RM_TYP 
			Season_Id 
			DY_TYP;
		
		if not 
	   	first.Room_Rate
		and not 
                last.Room_Rate 
then output;
	Run;

It didn't delete any observation within each group. Anything wrong with it? 

Thanks!

[PaigeMiller · Accepted solution]

use

FIRST.DY_TYP

and

LAST.DY_TYP

[Crubal]

This is awesome, thanks!