topic Re: Matching strings with spelling differences in script in QlikView

Matching strings with spelling differences in script

Mon, 02 Mar 2015 17:19:00 GMT

I am loading sales ndata. The customers names in these data are not always consistently spelled. Is there a way to automatically match nearly identically spelled customer names? Please note that I have thousands of names, so I cannot manually make a matching table.

Re: Matching strings with spelling differences in script

robert_mika — Mon, 02 Mar 2015 17:23:51 GMT

Maybe this will help

Fuzzy Search in Scripts | Qlik Community

Re: Matching strings with spelling differences in script

eespiritu — Mon, 02 Mar 2015 17:29:46 GMT

Hi Marc,

Does customer names data has an ID or something similar to identify each one of these customers (email)?

Re: Matching strings with spelling differences in script

Gysbert_Wassenaar — Mon, 02 Mar 2015 17:49:57 GMT

See this blog post: Teaching the load script new tricks

And my comment in the comments section.

Re: Matching strings with spelling differences in script

Mon, 02 Mar 2015 20:35:00 GMT

Not so. Otherwise, the solution would be trivial, isn't it,

Re: Matching strings with spelling differences in script

Mon, 02 Mar 2015 20:37:30 GMT

This is interesting. Not sure it will work for my issue. Some of the cuatomer names are not only spelled differntly, but also have some extra terms. E.g. Svenson and Svenson AB. Do you think these phonetic algorithms would work in this situation?

Re: Matching strings with spelling differences in script

Anonymous — Mon, 02 Mar 2015 21:08:52 GMT

Marc,

If there is no ID - there is no reliable solution. The solution to which gwassenaar is pointing is a solution for a different problem, that is how to merge similar sounding names into one. It may or may not be what you need.

The problem is that there is no way to tell that "Svenson" and "Svenson AB" are two different persons. To make the problem worse - you cannot tell for sure even that "Marc Donckers" and "Marc Donckers", spelled exactly the same way, is the same person. Can you? Are "Cassius Clay" and "Muhammad Ali" two different people? Not really. OK, the last example is rather uncommon, I admit...

So, you either have to create an ID combining several fields if possible (date of birth, eye color, etc.), or go manually.

Regards,

Michael