Fuzzy matching to consolidate company name

[sas028]

Hi,

I am trying to consolidate company names using fuzzy matching. I've attached sample data for different name my data currently has for Uber Eats, Uber Trip and Uber. 

I've created a fuzzy match workflow shown below and attached:

The output of the workflow is below: 

Is there something I could do in the fuzzy matching workflow so that supplier names where it say Uber Eats (UBER EATS 5KYJV, UBER EATS FTUDB) are grouped into UBER EATS and Uber Trip (UBER TRIP 26SAL, UBER TRIP 2AYSO) are grouped as UBER TRIP, within the fuzzy logic, without using a lookup table or formulas removing certain strings. 

[geraldo]

@sas028 

A WAY TO ANALYZE

if you improve the regex you won't need to replace

[sas028]

Hi Geraldo,

Thanks for your solution. The Uber companies are just a sample of the data I have. There are other companies with similar naming issues. Preferably, I'm looking for a solution using the fuzzy matching that will provide the naming convention I require without the use of formulas as I would have to do this for hundreds of different company names. 

Thanks! 

[jdminton]

Hey sas028,

I think you'd want to clean up the supplier names more before fuzzy matching. You can create a new field for supplier name and still retain the transactional information. I think most vendors will already have clean names, but if your data is coming from expense reports or some other transactional information, sometimes you get these extra codes included.

I'm struggling to come up with an automated way to do all at once, but you could create the new supplier column with cleaned up names for the handful of vendors with weird names that come through, it might be enough to do want you are wanting. If you set a record ID early in the process, you can also join back to the original data to see the other related details not part of the fuzzy match test.

I'm still trying to understand your end goal with the fuzzy match as it is definitely associating similar vendor names currently. Are you just trying to reduce the number of results to a manageable level?