topic Re: Identify/keep true consecutive negative value after positive value in SAS Programming

Identify/keep true consecutive negative value after positive value

Joshua017 — Tue, 17 Mar 2020 23:19:00 GMT

I have about 200,000 records of cases and I want to identify the true consecutive Negative result (can't have Positive result after it).

Here's some examples:

A: Keep only the result after first two positive (Obs 3-5)

B: Obs 6-7 need to be drop because these is a Positive after these Negative.

C: Need to be drop

😧 Only 15-17 can be keep since 13 have Positive after them as well.

The basic concept is to keep the Negative results after the last positive for each case (group?) and give them a seq# if possible.

Thank you so much for your help. I tried to add seq# but many times it starting with 2 or even 4 when I use

if first.Case then Seq+1;

Obs	Case	Test Date	Test Result	Record Wanted	Seq#
1	A	01/01/2020	Missing
2	A	01/02/2021	Positive
3	A	01/03/2021	Negative	V	1
4	A	01/04/2021	Negative	V	2
5	A	01/05/2021	Negative	V	3
6	B	01/01/2020	Negative
7	B	01/02/2021	Negative
8	B	01/03/2021	Positive
9	C	01/02/2021	Missing
10	D	01/01/2020	Positive
11	D	01/02/2021	Positive
12	D	01/03/2021	Missing
13	D	01/04/2021	Negative
14	D	01/01/2021	Positive
15	D	01/02/2022	Negative	V	1
16	D	01/03/2022	Negative	V	2
17	D	01/04/2022	Negative	V	3

Re: Identify/keep true consecutive negative value after positive value

Joshua017 — Tue, 17 Mar 2020 23:35:08 GMT

I got something like below but it's not working expected.

data Test_Negative;

set Tests;

by Case;

if Test_Result^="Negative" then T_Seq=0;

else if Test_Result="Negative" then T_Seq+1;

if first.Case & Test_Result^="Negative" then T_Seq=0;

else if first.Case & Test_Result="Negative" then T_Seq+1;

run;

Re: Identify/keep true consecutive negative value after positive value

Patrick — Wed, 18 Mar 2020 00:17:59 GMT

Welcome to the SAS Communities.

You need to read the source table twice because you can only figure out which rows to select once you've processed the last row within a group.

Below code should do what you're after.

data want;
  set have;
  by notsorted case;
  retain start_row n_row;
  if first.case then call missing(start_row, n_row);
  if Test_Result ne 'Negative' then call missing(start_row, n_row);
  else if missing(start_row) then start_row=_n_;
  if not missing(start_row) then n_row+1;
  if last.case and not missing(start_row) then 
    do;
      stop_row=start_row+n_row-1;
      do i=start_row to stop_row;
        set have point=i;
        output;
      end;
    end;
/*  drop start_row stop_row n_row i;*/
run;

proc print data=want;
run;

Re: Identify/keep true consecutive negative value after positive value

PGStats — Wed, 18 Mar 2020 03:47:10 GMT

A double do until() comes in handy for this task:

data have;
input Obs Case :$1. Date :mmddyy10. Result :$12.;
format date yymmdd10.;
datalines;
1   A   01/01/2020  Missing 　   　
2   A   01/02/2021  Positive    　   　
3   A   01/03/2021  Negative    V   1
4   A   01/04/2021  Negative    V   2
5   A   01/05/2021  Negative    V   3
6   B   01/01/2020  Negative         
7   B   01/02/2021  Negative         
8   B   01/03/2021  Positive         
9   C   01/02/2021  Missing 　   　
10  D   01/01/2020  Positive         
11  D   01/02/2021  Positive         
12  D   01/03/2021  Missing      
13  D   01/04/2021  Negative         
14  D   01/01/2021  Positive         
15  D   01/02/2022  Negative    V   1
16  D   01/03/2022  Negative    V   2
17  D   01/04/2022  Negative    V   3
;

data want;
do n = 1 by 1 until (last.case);
	set have; by case;
	if result = "Negative" then do;
		if missing(firstPos) then firstPos = n;
		lastPos = n;
		end;
	else call missing(firstPos);
	end;
seq = 0;
do n = 1 by 1 until (last.case);
	set have; by case;
	if firstPos > 0 then
		if lastPos > firstPos + 1 and n >= firstPos then do;
			seq + 1;
			output;
			end;
	end;
drop n firstPos lastPos;
run;

Re: Identify/keep true consecutive negative value after positive value

Ksharp — Wed, 18 Mar 2020 11:41:24 GMT

data have;
input Obs Case :$1. Date :mmddyy10. Result :$12.;
format date yymmdd10.;
datalines;
1   A   01/01/2020  Missing 　   　
2   A   01/02/2021  Positive    　   　
3   A   01/03/2021  Negative    V   1
4   A   01/04/2021  Negative    V   2
5   A   01/05/2021  Negative    V   3
6   B   01/01/2020  Negative         
7   B   01/02/2021  Negative         
8   B   01/03/2021  Positive         
9   C   01/02/2021  Missing 　   　
10  D   01/01/2020  Positive         
11  D   01/02/2021  Positive         
12  D   01/03/2021  Missing      
13  D   01/04/2021  Negative         
14  D   01/01/2021  Positive         
15  D   01/02/2022  Negative    V   1
16  D   01/03/2022  Negative    V   2
17  D   01/04/2022  Negative    V   3
;
data want;
do i=1 by 1 until(last.case);
 set have;
 by case ;
 if result='Positive' then  _i=i;
end;

do i=1 by 1 until(last.case);
 set have;
 by case result notsorted;
 if first.result then count=.;
 count+1;
 if i le _i or result ne 'Negative' then call missing(count);
 output;
end;
drop i _i;
run;