topic Re: How to find a list of specific strings amid a character variable? in SAS Procedures

How to find a list of specific strings amid a character variable?

Altal — Sat, 07 Sep 2013 17:44:54 GMT

I want to search a variable in a very big data set for a list of specific substring.

I though index function would do the job, but index function can fit only one string to search for.

I var on variable, and I want to search this variable for big number of small strings for example "DEX", HEX", "RED", "BET" and so on.

so I want any word in the search variable to be flagged if int has for example dexamethsone, hexagonal...etc

I was thinking this would work but it's not

if index(VAR, 'DEX', 'HEX', 'BET') >0 then flag=1

I don't want to specify many if statements, because my st of search is big

Any ideas.

Thanks

Re: How to find a list of specific strings amid a character variable?

Astounding — Sat, 07 Sep 2013 18:20:52 GMT

There's no way I know of to search for all three at the same time. To cut down the costs a bit, you can use:

if index(VAR, 'DEX') then flag=1;

else if index(VAR, 'HEX') then flag=1;

else if index(VAR, 'BET') then flag=1;

Good luck.

Re: How to find a list of specific strings amid a character variable?

Haikuo — Sun, 08 Sep 2013 00:54:13 GMT

Maybe some PRX?

FLAG=(prxmatch('/[DEX|HEX|RED|BET]/',VAR)>0);

Haikuo

Re: How to find a list of specific strings amid a character variable?

Altal — Sun, 08 Sep 2013 03:41:36 GMT

Thank you guys. That was really helpful.

Haikuo- thank you, I didn't think about the prxmatch function before. It worked; I minimally modified your code to work. Here is what worked for me...

FLAG=(prxmatch('/DEX|HEX|RED|BET/',VAR)>0);

You can add any substring and divide them by "|". This will save from putting multiple IF statements if looking for various substrings.

Re: How to find a list of specific strings amid a character variable?

PGStats — Sun, 08 Sep 2013 03:52:58 GMT

You might want to add a suffix to the pattern to improve efficiency and behaviour:

FLAG=(prxmatch('/DEX|HEX|RED|BET/io',VAR)>0);

The 'i' makes the match case insensitive and the 'o' instructs the function that the pattern should be compiled only once, otherwise, it is recompiled on every call.

Re: How to find a list of specific strings amid a character variable?

Altal — Sun, 08 Sep 2013 04:13:55 GMT

Yep- worked like magic! Thanks PGStats