topic Re: categorize variable based on matching words in SAS Programming

categorize variable based on matching words

eemrun — Wed, 28 Mar 2018 01:58:37 GMT

I am trying to create an expense report where I am categorizing transactions based on certain keywords (e.g. if a transaction line contains STARBUCKS, categorize it as EATING-OUT). I am using the following but notice that the command does not always work (as in does not assign transactions to categories all the time. Is there a better way to do this other than the following?

	if prxmatch ("m/BAKE & BEANS|Smart Sushi |Columbus Coffee|Coffee Club|Mozaic|Menulog|Kyoto|Baxters|
		NZ Natural|Rio|yoway|yoghurt||Dogmatic|Donut|Thanya|Nol|Bean/i",details) > 0 then
		Category = "Eat-Out";

Re: categorize variable based on matching words

ChrisNZ — Wed, 28 Mar 2018 02:43:23 GMT

There is no reason this should fail if done properly.

Do you have an example of data that is missed?

Note that you have a double pipe || in the example you gave.

Or... Could it be that you look for spaces when you shouldn't? Try:

if prxmatch ("m/BAKE & BEANS|Smart Sushi |Columbus Coffee|Coffee Club|Mozaic|Menulog|Kyoto|Baxters|"
           ||"NZ Natural|Rio|yoway|yoghurt|Dogmatic|Donut|Thanya|Nol|Bean/i",details
            ) > 0 then category = "Eat-Out";

Re: categorize variable based on matching words

eemrun — Thu, 29 Mar 2018 02:49:36 GMT

Thanks. You are correct. I had some small syntax errors and some classification errors. All works now.