Solved: Retain and getting max values

RAVI2000 · Posted 07-09-2021 08:09 PM

I have the below data. I want to create new variables with last aedtc date with corresponding max grade and where ever aeevnt is 1 I want to retain to the corresponding record.

data data;
input @1 seqno $8.	@9 grade 1. @11 aedtc $11. @22 aeevnt 1.;
cards;
K-001   1 .          0
K-001   2 .          0
K-001   2 19DEC2019  1
K-001   3 19DEC2019  1
K-001   3 .          0
K-002   1 .          0
K-002   4 .          0
K-002   3 20APR2020  1
K-002   1 .	         0
K-003   2 20JAN2020	 1
K-003   2 20JAN2020	 1
K-003   1 .	         0
K-003   3 .	         0
K-003   3 23JAN2020	 1
K-003   3 .	         0
K-003   1 .	         0
K-003   . .	         0
K-003   1 .	         0
K-003   3 .	         0
K-003   2 .	         0
K-005   2 .	         0
K-005   1 .	         0
K-005   1 .	         0
K-006   3 30APR2021	 1
K-006   3 30APR2021	 1
K-006   1 .	         0
K-006   2 06APR2021	 1
K-006   2 06APR2021	 1
K-006   1 .	         0
K-006   2 .	         0
K-006   3 .	         0
K-006   1 .	         0
K-007   1 .	         0
K-007   3 .	         0
K-007   1 .	         0
K-007   1 .	         0
DK-001  2 .	         0
DK-001  . .	         0
DK-001  2 .	         0
DK-001  . .	         0
DK-001  1 .	         0
DK-003  . .	         0
DK-003  2 .	         0
DK-003  . .	         0
DK-003  2 .	         0
DK-003  1 .	         0
DK-003  1 .	         0
DK-005  . .	         0
DK-006  . .	         0
DK-007  . .	         0
CDK-001 2 .	         0
CDK-001 3 .	         0
CDK-001 2 .	         0
CDK-001 3 2 4SEP2020 1
CDK-001 3 .	         0
CDK-002 1 .	         0
CDK-002 3 .	         0
;
run;

My final output dataset should have only one record per pateint.
K-001 3 19DEC2019 1
K-002 3 20APR2020 1
K-003 3 23JAN2020 1
K-005 2 . 0
K-006 3 06APR2021 1
K-007 3 . 0
DK-001 2 . 0
DK-003 2 . 0
DK-005 . . 0
DK-006 . . 0
DK-007 . . 0
CDK-001 3 2 4SEP2020 1
CDK-002 3 . 0

mkeintz · Posted 07-10-2021 01:13 AM

@RAVI2000 wrote:
If you observe the data for grade variable with subject K-002 has max grade 4. It's not capturing the max grade here. I tried modifying but still not working.

The reason @novinosrin 's program does not generate a max grade 4 for K-002 is because neither your sample solution data, nor does your description suggest it should generate a 4. The 4 only occurs with a missing aedtc date. Your description presumably would rule that out:

"last aedtc date with corresponding max grade"

Here is a data step solution that retains the last record with a date (not necessarily the latest date), and retains the max grade for dated, and max grade for undated records. It then generates the results you specify:


data want (drop=_:);
  set data;
  by seqno notsorted;

  retain _lastest_date _maxgrade_dated  _maxgrade_undated ;
  if first.seqno then call missing (of _:);

  if aedtc^=. then do;
    _lastest_date=aedtc;
    _maxgrade_dated=max(_maxgrade_dated,grade);
  end;
  else if grade^=. then _maxgrade_undated=max(_maxgrade_undated,grade);

  if last.seqno;
  aedtc=_lastest_date;
  grade=coalesce(_maxgrade_dated,_maxgrade_undated);
  if aedtc^=. then aeevnt=1;
run;

--------------------------
The hash OUTPUT method will overwrite a SAS data set, but not append. That can be costly. Consider voting for Add a HASH object method which would append a hash object to an existing SAS data set

Would enabling PROC SORT to simultaneously output multiple datasets be useful? Then vote for
Allow PROC SORT to output multiple datasets

--------------------------

View solution in original post

tarheel13 · Posted 07-09-2021 08:28 PM

data data;
input @1 seqno $8.	@9 grade 1. @11 aedtc $11. @22 aeevnt 1.;
cards;
K-001   1 .          0
K-001   2 .          0
K-001   2 19DEC2019  1
K-001   3 19DEC2019  1
K-001   3 .          0
K-002   1 .          0
K-002   4 .          0
K-002   3 20APR2020  1
K-002   1 .	         0
K-003   2 20JAN2020	 1
K-003   2 20JAN2020	 1
K-003   1 .	         0
K-003   3 .	         0
K-003   3 23JAN2020	 1
K-003   3 .	         0
K-003   1 .	         0
K-003   . .	         0
K-003   1 .	         0
K-003   3 .	         0
K-003   2 .	         0
K-005   2 .	         0
K-005   1 .	         0
K-005   1 .	         0
K-006   3 30APR2021	 1
K-006   3 30APR2021	 1
K-006   1 .	         0
K-006   2 06APR2021	 1
K-006   2 06APR2021	 1
K-006   1 .	         0
K-006   2 .	         0
K-006   3 .	         0
K-006   1 .	         0
K-007   1 .	         0
K-007   3 .	         0
K-007   1 .	         0
K-007   1 .	         0
DK-001  2 .	         0
DK-001  . .	         0
DK-001  2 .	         0
DK-001  . .	         0
DK-001  1 .	         0
DK-003  . .	         0
DK-003  2 .	         0
DK-003  . .	         0
DK-003  2 .	         0
DK-003  1 .	         0
DK-003  1 .	         0
DK-005  . .	         0
DK-006  . .	         0
DK-007  . .	         0
CDK-001 2 .	         0
CDK-001 3 .	         0
CDK-001 2 .	         0
CDK-001 3 04SEP2020  1
CDK-001 3 .	         0
CDK-002 1 .	         0
CDK-002 3 .	         0
;
run;

proc sql number;
	create table data2 as select * from data
		group by seqno
		having grade=max(grade)
		order by seqno, aedtc;
quit;

data data3;
	set data2;
	by seqno aedtc;
	if last.seqno then output;
run;

RAVI2000 · Posted 07-09-2021 11:06 PM

It works!! Can this also be explained in first. last. concept ?

tarheel13 · Posted 07-10-2021 04:39 AM

If it’s the first seqno in a by group then first.seqno =1. Otherwise it’s 0. If it’s the last seqno in a group then last.seqno =1. You only wanted 1 row per subject so I did if last.seqno then output.

novinosrin · Posted 07-09-2021 10:42 PM

Using the date variable aedtc as numeric-

data data;
input @1 seqno $8.	@9 grade 1.  aedtc :date9. @22 aeevnt 1.;
format aedtc date9.;
cards;
K-001   1 .          0
K-001   2 .          0
K-001   2 19DEC2019  1
K-001   3 19DEC2019  1
K-001   3 .          0
K-002   1 .          0
K-002   4 .          0
K-002   3 20APR2020  1
K-002   1 .	         0
K-003   2 20JAN2020	 1
K-003   2 20JAN2020	 1
K-003   1 .	         0
K-003   3 .	         0
K-003   3 23JAN2020	 1
K-003   3 .	         0
K-003   1 .	         0
K-003   . .	         0
K-003   1 .	         0
K-003   3 .	         0
K-003   2 .	         0
K-005   2 .	         0
K-005   1 .	         0
K-005   1 .	         0
K-006   3 30APR2021	 1
K-006   3 30APR2021	 1
K-006   1 .	         0
K-006   2 06APR2021	 1
K-006   2 06APR2021	 1
K-006   1 .	         0
K-006   2 .	         0
K-006   3 .	         0
K-006   1 .	         0
K-007   1 .	         0
K-007   3 .	         0
K-007   1 .	         0
K-007   1 .	         0
DK-001  2 .	         0
DK-001  . .	         0
DK-001  2 .	         0
DK-001  . .	         0
DK-001  1 .	         0
DK-003  . .	         0
DK-003  2 .	         0
DK-003  . .	         0
DK-003  2 .	         0
DK-003  1 .	         0
DK-003  1 .	         0
DK-005  . .	         0
DK-006  . .	         0
DK-007  . .	         0
CDK-001 2 .	         0
CDK-001 3 .	         0
CDK-001 2 .	         0
CDK-001 3 04SEP2020  1
CDK-001 3 .	         0
CDK-002 1 .	         0
CDK-002 3 .	         0
;
run;

proc sql;
  create table want as
  select distinct *
  from data 
  group by seqno
  having aeevnt and max(grade*aeevnt)=grade
  or not max(aeevnt) and max(grade)=grade;
 quit;


seqno	grade	aedtc	aeevnt
CDK-001	3	04SEP2020	1
CDK-002	3	.	0
DK-001	2	.	0
DK-003	2	.	0
DK-005	.	.	0
DK-006	.	.	0
DK-007	.	.	0
K-001	3	19DEC2019	1
K-002	3	20APR2020	1
K-003	3	23JAN2020	1
K-005	2	.	0
K-006	3	30APR2021	1
K-007	3	.	0

RAVI2000 · Posted 07-09-2021 11:20 PM

If you observe the data for grade variable with subject K-002 has max grade 4. It's not capturing the max grade here. I tried modifying but still not working.

mkeintz · Posted 07-10-2021 01:13 AM

@RAVI2000 wrote:
If you observe the data for grade variable with subject K-002 has max grade 4. It's not capturing the max grade here. I tried modifying but still not working.

The reason @novinosrin 's program does not generate a max grade 4 for K-002 is because neither your sample solution data, nor does your description suggest it should generate a 4. The 4 only occurs with a missing aedtc date. Your description presumably would rule that out:

"last aedtc date with corresponding max grade"

Here is a data step solution that retains the last record with a date (not necessarily the latest date), and retains the max grade for dated, and max grade for undated records. It then generates the results you specify:


data want (drop=_:);
  set data;
  by seqno notsorted;

  retain _lastest_date _maxgrade_dated  _maxgrade_undated ;
  if first.seqno then call missing (of _:);

  if aedtc^=. then do;
    _lastest_date=aedtc;
    _maxgrade_dated=max(_maxgrade_dated,grade);
  end;
  else if grade^=. then _maxgrade_undated=max(_maxgrade_undated,grade);

  if last.seqno;
  aedtc=_lastest_date;
  grade=coalesce(_maxgrade_dated,_maxgrade_undated);
  if aedtc^=. then aeevnt=1;
run;

--------------------------
The hash OUTPUT method will overwrite a SAS data set, but not append. That can be costly. Consider voting for Add a HASH object method which would append a hash object to an existing SAS data set

Would enabling PROC SORT to simultaneously output multiple datasets be useful? Then vote for
Allow PROC SORT to output multiple datasets

--------------------------

Retain and getting max values

Re: Retain and getting max values

Re: Retain and getting max values

Re: Retain and getting max values

Re: Retain and getting max values

Re: Retain and getting max values

Re: Retain and getting max values

Re: Retain and getting max values

Classroom Training Available!