topic Re: string match (whole or partial) in SAS Programming

string match (whole or partial)

SASPhile — Mon, 12 Feb 2018 21:22:24 GMT

Need to compare two columns either complete or partial. This can be achived using regex,but do not want to hardcode the values as I need to compare 100's of records.

for instance:

ColA ColB

AE AE_NCI

AE CM_AE

AE SAE_SUPP

the colA has the string "AE" , if Colb has complete or partial that assign value as 1 . There are 100's of rows with different values that need to compared.

usual way is if prxmatch("/ae/oi",colb)>0 then value=1. but here ae is hardcoded. anyway to dymaically pass colA into prxmatch function?

Re: string match (whole or partial)

Reeza — Mon, 12 Feb 2018 21:29:21 GMT

FINDW or INDEXW?

Re: string match (whole or partial)

SASPhile — Mon, 12 Feb 2018 21:31:27 GMT

SAE_SUPP is not resulted out

Re: string match (whole or partial)

art297 — Mon, 12 Feb 2018 21:34:33 GMT

How about:

data have;
  input (ColA    ColB) ($);
  cards;
AE       AE_NCI
AE       CM_AE
AE       SAE_SUPP
AE       AB_NCI
AB       AB_NCI
AB       CM_AE
AB       SAB_SUPP
;

data want;
  set have;
  if find(ColB,ColA,'it') then value=1;
  else value=0;
run;

Art, CEO, AnalystFinder.com

Re: string match (whole or partial)

SASPhile — Mon, 12 Feb 2018 21:45:53 GMT

Artc, I'm getting the value 0 for all records

Re: string match (whole or partial)

Reeza — Mon, 12 Feb 2018 21:49:56 GMT

I ran @art297 code and it worked fine with the demo data. Does it work with the demo data he shows? Does it not work with your actual data instead. Also, Art is correct, if you want partial searchs,use FIND, if searching for a specific character it's FINDC, and for a word it's FINDW.

Re: string match (whole or partial)

SuryaKiran — Mon, 12 Feb 2018 21:53:11 GMT

find(ColB,STRIP(ColA),'it')

Remove leading and trailing blanks.

Re: string match (whole or partial)

art297 — Mon, 12 Feb 2018 22:22:19 GMT

I agree with @SuryaKiran. If your colA includes leading and trailing blanks, then use the strip function. However, then remove the t option as it will only cause the process to slow down. i.e.:

data want;
  set have;
  if find(ColB,strip(ColA),'i') then value=1;
  else value=0;
run;

Art, CEO, AnalystFinder.com

p.s., @SASPhile: This is ArtT not ArtC

Re: string match (whole or partial)

ChrisNZ — Tue, 13 Feb 2018 02:29:17 GMT

Like this?

data HAVE;
  input COLA $ COLB $;
  A= prxmatch(catt('/(',COLA,')/i'),COLB);
  putlog A=;
cards;
AE AE_NCI
AE CM_AE
AE SAE_SUPP
AX SAE_SUPP
run;

A=1
A=4
A=2
A=0

RegEx are more expensive than functions like FIND though.