topic Fuzzy find/kindex in SAS Programming

Fuzzy find/kindex

daradanye — Thu, 07 Mar 2019 20:40:25 GMT

Hi,

I have a problem about fuzzy find/kindex.

I have two columns, "All_members" and "Excellent". I want to generate a new column "Excellent_in" to indicate if "Excellent" is in "All_members". I usually do this by writing:

if kindex(All_members, Excellent)>0;

However, as it is shown, the name does not perfectly match here. I guess I need some fuzzy find/kindex function here. I will appreciate it very much if someone can help me with this. Thanks!

All_members	Excellent	Excellent_in
Lee Han llp, Det tet co., 3M Inc.	3M	1
……	……	……
Vans co, nineng inc., leiwo	wyu	0
Vans co, nineng inc., leiwo	leiwo inc	1

Re: Fuzzy find/kindex

ChrisNZ — Thu, 07 Mar 2019 22:38:05 GMT

kindex() should not be fuzzy.

I could not replicate from your example sadly.

data _null_;
  A=kindex('Vans co, nineng inc., leiwo', 'leiwo inc');  
  put A=; 
run;

A=0

Re: Fuzzy find/kindex

daradanye — Fri, 08 Mar 2019 15:38:30 GMT

Thanks, ChrisNZ!

Yes, I understand that kindex could not be fuzzy. I am asking if there are some functions that can be used to fuzzy matching.

Re: Fuzzy find/kindex

ChrisNZ — Sun, 10 Mar 2019 04:10:45 GMT

> if there are some functions that can be used to fuzzy matching

Yes, but you have to define the rules yourself.

In the example you give, you could try to simply use index() on strings that have been cleaned: uppercase, no punctuation, no "Ltd" "Inc" "Limited" etc.

Functions that do fuzzy matching include the like and sounds like operators in SQL, the regular expressions matches like function prxmatch(), and the spelling distance functions like spedis() or compged(). It is up to you to define the criteria though. Each case is different.