topic Re: String Difference in Percentage in SAS Procedures

String Difference in Percentage

JAR — Sun, 06 Jul 2014 17:55:18 GMT

Dear All,

I would like to learn how to calculate % change, when two strings are compared.

Here is the original file:

1	RAT	CAT
2	BELL	BALL
3	TIME	TOM

As there is only one character that is different in the first observation, it should return 1/3 %(number of deviation / total number of char in the original).

Desired output is something like this:

1	RAT	CAT	33
2	BELL	BALL	25
3	TIME	TOM	50

Thanks in advance,

Jijil Ramakrishnan

Re: String Difference in Percentage

stat_sas — Sun, 06 Jul 2014 18:51:52 GMT

Try this for the desired output.

data have;
input obs var1 $ var2 $;
datalines;
1 RAT CAT
2 BELL BALL
3 TIME TOM
;

data want;
set have;

do i=1 by 1 while(substr(var1,i,1) ne ' ');
if index(var2,substr(var1,i,1))>0 then cnt+1;
output;
end;
cnt=0;
run;

data final (drop=i cnt);
set want;
by obs;
if last.obs;
diff=(i-cnt)/i*100;
format diff 8.0;
run;

Re: String Difference in Percentage

PGStats — Sun, 06 Jul 2014 19:24:09 GMT

Approximate matches between strings is an already well researched topic. SAS provides you with many tools in that area. Please look at SAS functions COMPGED, COMPLEV, SOUNDEX, and SPEDIS.

Re: String Difference in Percentage

JAR — Tue, 08 Jul 2014 19:38:16 GMT

Thank you so much.

Re: String Difference in Percentage

JAR — Tue, 08 Jul 2014 19:39:32 GMT

Thanks a lot!