topic Re: Do Statement in SAS Programming

Split records using DO Statement

Santhosh01382 — Tue, 21 Sep 2021 06:23:54 GMT

usubjid ARMCD ARM
7959011 GPRTNKYKNSLWPCJEHGKG ACCUVE1\ACCUVE2\ACCUVE3\ACCUVE4\ACCUVE5\ACCUVE6\ACCUVE7\ACCUVE8\ACCUVE9\ACCUVE10

I want to split above single record into 10 records and each record should contains subjid, armcd and arm and each record should populate respected ARM and ARMCD accordingly.
By using the delimeter '\' successfully I have spatted ARM variable into 10 records but I am unable to split ARMCD accordingly with below program. Can anyone please help on this.
ARMCD has 20 characters and each 2 chars are belongs to 1 treatment.
GP = ACCUVE1
RT = ACCUVE2
NK = ACCUVE3.....

In final dataset I need
usubjid ARMCD ARM
7959011 GP ACCUVE1
7959011 RT ACCUVE2
7959011 NK ACCUVE3................

Thanks in advance.

MY PROGRAM:
%let p = 0;

data trtp_bi_test;
set t58;

do i = 1 to 10;
length trtpval $200;
trtp = "TRT"||put(i,z2.)||"P";
trtplbl = "Planned Treatment for Period "||put(i,z2.);
trtpval = scan(armf,i,"\");

j=0;
trtpcdval = substr(armcd, i+j, 2);
j+1;
output;
end;
run;

Re: Do Statement

maguiremq — Thu, 16 Sep 2021 23:49:38 GMT

I kept trying to do this in one DATA step, but I had to break it down into two separate ones. I know someone else would know how to do it in one.

data have;
input usubjid 8. armcd :$30. arm :$300.;
infile datalines delimiter = " ";
datalines;
7959011 GPRTNKYKNSLWPCJEHGKG ACCUVE1\ACCUVE2\ACCUVE3\ACCUVE4\ACCUVE5\ACCUVE6\ACCUVE7\ACCUVE8\ACCUVE9\ACCUVE10
;
run;

data armcd (drop = i armcd max_len);
	set have (keep = usubjid armcd);
	max_len = max(length(armcd));
	do i = 1 to length(armcd) by 2 until (i > max_len);
		armcd_t = substr(armcd, i, 2);
		output;
	end;
run;

data arm (drop = i arm);
	set have (keep = usubjid arm);
	do i = 1 to countw(arm, "\");
		arm_t = scan(arm, i, "\");
		output;
	end;
run;

data want;
	merge
			armcd 	(in = a rename = (armcd_t = armcd))
			arm 	(in = b rename = (arm_t = arm));
	by		usubjid;
			if a and b;
run;

Obs 	usubjid 	armcd 	arm
1 	7959011 	GP 	ACCUVE1
2 	7959011 	RT 	ACCUVE2
3 	7959011 	NK 	ACCUVE3
4 	7959011 	YK 	ACCUVE4
5 	7959011 	NS 	ACCUVE5
6 	7959011 	LW 	ACCUVE6
7 	7959011 	PC 	ACCUVE7
8 	7959011 	JE 	ACCUVE8
9 	7959011 	HG 	ACCUVE9
10 	7959011 	KG 	ACCUVE10

Re: Do Statement

Tom — Fri, 17 Sep 2021 00:53:08 GMT

Just use a little arithmetic to figure out where each ARMCD starts.

data have;
  input usubjid armcd :$30. arm :$300.;
datalines;
7959011
GPRTNKYKNSLWPCJEHGKG
ACCUVE1\ACCUVE2\ACCUVE3\ACCUVE4\ACCUVE5\ACCUVE6\ACCUVE7\ACCUVE8\ACCUVE9\ACCUVE10
;

data want; 
  set have(rename=(armcd=armcd_string arm=arm_string));
  length ARMCD $2 ARM $20;
  do index=1 to 10 ;
    ARMCD=substr(armcd_string,2*(index-1)+1,2);
    ARM=scan(arm_string,index,'\');
    output;
  end;
  drop armcd_string arm_string;
run;

Obs    usubjid    ARMCD      ARM       index

  1    7959011     GP      ACCUVE1        1
  2    7959011     RT      ACCUVE2        2
  3    7959011     NK      ACCUVE3        3
  4    7959011     YK      ACCUVE4        4
  5    7959011     NS      ACCUVE5        5
  6    7959011     LW      ACCUVE6        6
  7    7959011     PC      ACCUVE7        7
  8    7959011     JE      ACCUVE8        8
  9    7959011     HG      ACCUVE9        9
 10    7959011     KG      ACCUVE10      10