Select codes that begins with tree alphabetic characters

[Chris_LK_87]

Hello,

I have dataset that looks like this. Each line contains an individual and codes attached to that individual. The codes are stored with spaces in one variable. 

data have;
length id $10 dcode $48;
input id$ dcode$ &;
datalines;
1 MCB10 PC001 AAA30
2  AC003 PA000 TAC25
3  QC000 CAB50 FCE10
;
run;

I would like to select every code that begins with tree alphabetic characters and place each code in a new variable. 

Data want; 

set have;

id$ dcode1$ dcode2$ dcode3$

1  MCB10             AAA30

2                          TAC25

3              CAB50 FCE10

[Reeza · Accepted solution]

1. Use COUNTW() to count number of words

2. Extract each word

3. Check if first 3 characters are alphabetic (NOTALPHA())

4. Output if valid record

5. Transpose to a wide structure as desired

data have;
	length id $10 dcode $48;
	input id$ dcode$ &;
	datalines;
1 MCB10 PC001 AAA30
2  AC003 PA000 TAC25
3  QC000 CAB50 FCE10
;
run;

data long;
	set have;
	nwords=countw(dcode); /*1*/

	do i=1 to nwords;
		word=scan(dcode, i); /*2*/

		if not notalpha(substr(word, 1, 3)) then /*3*/
			output; /*4*/
	end;
run;

/*5*/
proc transpose data=long out=want prefix=WORD_;
	by ID;
	var word;
run;

---

@Chris_LK_87 wrote:

Hello,

I have dataset that looks like this. Each line contains an individual and codes attached to that individual. The codes are stored with spaces in one variable. 

data have;
length id $10 dcode $48;
input id$ dcode$ &;
datalines;
1 MCB10 PC001 AAA30
2  AC003 PA000 TAC25
3  QC000 CAB50 FCE10
;
run;

I would like to select every code that begins with tree alphabetic characters and place each code in a new variable. 

Data want; 

set have;

id$ dcode1$ dcode2$ dcode3$

1  MCB10             AAA30

2                          TAC25

3              CAB50 FCE10

---

[Chris_LK_87]

Thanks!