Fuzzy Matching

[ehellblom]

Hi all,

I am semi-new to the Fuzzy Match tool and I am trying to match misspelled words, not the usual Name and Address a lot of tutorials are about so been struggling to sort it out.

My main table is comes from a "Free text" field where all types of fat-fingers can happen so trying to match it to a Keyword table so they all have the same spelling. The main table itself (and the real table) has been created by splitting all the words out to rows and in capitals to identify each word and then it will be joined back but first I need to get the Fuzzy Matching right.

I have attached a workflow and I have tried to set it up with the main table with all types of different spellings and then one with the keywords I would like to match them with (how they are supposed to be spelled). Some are correct but in plural and I can't get them to match them either. 

Could someone please help to identify where I go wrong/what are the recommended settings or maybe it is not possible to do on single words, just names and addresses? If so, is there another way to do it? My goal is to avoid doing a long IF statement trying to capture all the spellings.

Thank you!

[fmvizcaino · Accepted solution]

Hi @ehellblom ,

Attached is a suggestion for you to make in your fuzzy match workflow.

I've changed:

1 fuzzy match method from metaphones to whole field - since it is mistyping error,

2 key length to be 1 and inserted a dummy character - so it will consider everything since all lines will have the same dummy key for starters

3 I've inserted the make group tool to group everything as it should - it works perfectly with the recordID because it orders everything correctly as you might see in the example I've built.

Best,

Fernando V.

[ehellblom]

@fmvizcaino - that is amazing! Thank you much, with a few tweaks it worked great with my real data!

[IJK]

Would love to know what your tweaks were!

[mihir_mir_jb]

@fmvizcaino I am trying to do something similar with my data, where I have company names that are tagged differently like in the raw data file named 'Company Cluster' where I am trying to cluster similar names together. Though I am not getting the output as expected, for instance I would like to cluster first 5 companies together named ABC into one name may be ABC Company Limited which is currently not happening. I have attached the workflow and the output file named 'Alteryx Fuzzy Logic output' 

Could you please help me to guide how can go about this.