Fuzzy Match on Nicknames not working as expected

Hello Everyone!

I am trying to demonstrate to some colleagues that Alteryx is the best tool for Fuzzy Matching (as opposed to Kalido MDM, Oracle Text, or others). They are very concerned about Thesaurus matching (the examples they use are matching "William" with "Bill" for names or "Drive" with "Dr" for addresses). I decided to test these specific examples and to my dismay, they FAILED!

So I took a step back and looked at the Fuzzy Match tool documentation:

Excerpt from Fuzzy Match Documentation.PNG

I decided to test fuzzy matching with Andrew, Andy, and Drew using the Name with Nicknames match style as suggested....

Sample Name Data.PNG Fuzzy Match Nickname Configuration.PNG

....and Alteryx returned NO MATCHES!!!!

I've attached a workflow showing my failed attempt to match Andrew, Drew, and Andy or Bill with William. I also have there my failed attempt to match "333 Reed Drive" with "333 Reed Dr" with the Address No Suite algorithm.

Please help me figure out how I can get Alteryx to see these as matches! (I have to be missing something if the example in the process documentation doesn't even work!) Also, does anyone know where Alteryx stores the "common Nicknames table" mentioned in the documentation?

Thank you!

P.S. I have already tried lowering the match strength; still no success!

Fuzzy Match Mystery.yxmd

Best Practices

Fuzzy Match

Accepted answers

RodL

As to the location for the nicknames file, in a default install it should be located in...

C:\Program Files\Alteryx\bin\RuntimeData\FuzzyMatch\Nicknames

Note that you can add your own YXDB in that location and the dropdown should pick up the additional files to use.

cbridges

I'm following this post with great interest, because fuzzy matching has always given me a fuzzy head!

There's an Alteryx webinar on fuzzy matching next Tuesday... http://www.alteryx.com/virtual-training... and I'll be there for sure!

All comments

RodL

As to the location for the nicknames file, in a default install it should be located in...

C:\Program Files\Alteryx\bin\RuntimeData\FuzzyMatch\Nicknames

Note that you can add your own YXDB in that location and the dropdown should pick up the additional files to use.

DultonM

Great to know Rod! Thank you!

It is interesting/frustrating that Andrew --> Andy and Andrew--> Drew are in that Nicknames table...but they don't match in my example....why is that?

DultonM

Update: @patrick_digan worked with me offline and discovered that if you edit the Names with Nicknames option to NOT generate keys, all of a sudden Andrew matches with Andy and Drew!! (Note, at least 1 field has to generate keys so we included a dummy field where we gave every record the same last name and chose match style = Exact for that field). Other nicknames worked as well! My conclusion is that the presence of the keys on the Names w/ Nicknames option prevents these Nicknames from matching despite the alleged use of the common nicknames table! Is this an appropriate conclusion? (I hope not because that is the same thing as saying the Nicknames table is not even applied!)

cbridges

I'm following this post with great interest, because fuzzy matching has always given me a fuzzy head!

There's an Alteryx webinar on fuzzy matching next Tuesday... http://www.alteryx.com/virtual-training... and I'll be there for sure!

Quick Links

This months top contributors

atcodedog05 19458

Qiu 15866

binu_acs 15708

MarqueeCrew 13708

apathetichell 13703