Lag Function

[Aidan]

Hi there

I want to use the lag function to return the final column on the right of the attached.

When the month changes I need the lag to reset itself.

Has anyone any helpful code so I can achieve this?

Thanks,
Aidan 

01-Oct-15	1	196684	187176	1916711	10.2401537	2.02054
01-Oct-15	2	1916711	1660708	3355534.07	2.02054429	1.24276
01-Oct-15	3	3355534.07	2976006	3698454	1.24275755	1.25583
01-Oct-15	4	3698454	3039553	3817150.07	1.25582613	0.93742
01-Oct-15	5	3817150.07	2152312	2017629.07	0.93742407	1.21171
01-Oct-15	6	2017629.07	894441	1083806	1.21171314	1.31364
01-Oct-15	7	1083806	518112	680615	1.31364454	1
01-Oct-15	8	680615	680615	.	1	1
01-Nov-15	1	196684	187176	196684	1.05079711	1.15415
01-Nov-15	2	1916711	1660708	1916711	1.15415293	1.12753
01-Nov-15	3	3355534.07	2976006	3355534.07	1.12752931	1.21678
01-Nov-15	4	3698454	3039553	3698454	1.21677562	1.77351
01-Nov-15	5	3817150.07	2152312	3817150.07	1.77351149	2.25574
01-Nov-15	6	2017629.07	894441	2017629.07	2.25574287	1
01-Nov-15	7	1083806	1083806	1083806	1	1

[Tom · Accepted solution]

The last column on the right looks like it is actually the data from the next row and not the previous row that LAG() would allow. There is no LEAD() function so you will probably want read the input data twice.

You didn't provide variable names so lets assume that the first column is called MONTH and the others are VAR2 to VAR7.

data want ;
  set have end=eof;
  by MONTH;
  if not eof then set have(keep=VAR6 rename=(VAR6=VAR7) firstobs=2);
  if last.MONTH then VAR7=1 ;
run;

[Aidan]

Thanks for the reply

My code is as follows with the updated Variables

Do I need to have an input statement before your code also?

data want ;
set have end=eof;
by MEM_TED;
if not eof then set have(keep=INC_GRUF rename=(INC_GRUF=NEXT_INC_GRUF) firstobs=2);
if last.MEM_TED then NEXT_INC_GRUF=1 ;
run;

[Tom]

You do not need an input statement as you need to have a SAS dataset and not just a raw text file to run that code.  If you do not yet have the data in a dataset then run a datastep before hand to create it.  You should probably also change the dataset names of HAVE and WANT to match your actual dataset names.

Also your data will need to be sorted by MEM_TED and at least one other variable to make sure that it is is in the right order.  You might want to add a PROC SORT step to make sure if you are not positive that the data is sorted properly.

Also note that your MEM_TED variable looks like a date.  If you want to group by that variable make sure that all of the records use the same day of the month (in your example they all use the first day of the month).  Otherwise the different dates will be treated as separate by groups.