Fuzzy Match is creating duplicate records

[monish_chandra]

I'm trying to match a list of ID, CPT fields which will be Exact Match and Description field as Custom in Fuzzy Match tool. The description consists of Alphabets and digits. 

I've currently set the Match Threshold to 50%. I tried to use 'Character: Best of Jaro and levenshtein distance' to check each character in the description, but it is not working as expected. In the attached workflow I have set the Match function to : Words Jaro Distance. Can you please check if the settings are correct for this scenario?

Input

RecordID	ID	CPT	DESCRIPTION	Location	FAC1	FAC2	FAC3	FAC4	FAC5	FAC6	FAC7
1	378	G3857	DEHYDRATION INFUS 1ST HOUR	ABC	0	62	0	0	0	34	12
2	378	G3857	IV INFUSN DEHYDRATION INITIAL HR	MTB	235	0	377	888	0	0	0
3	293	G3857	DEHYDRATION INFUS 1ST HOUR	ABC	0	765	0	0	328	371	278

Record 1 and 2 have the same ID and same CPT. The description also matches upto 50%. Can I have the Description for ABC and Description for MTB side by side and summarize the FACs individually. I tried to do that in the attached workflow but the Fuzzy match tool is creating so many duplicates. As per my understanding, it takes each description and looks for a similar description. In the next iteration if compares records that are already compared.  

I'd like to see the Matched records and Unmatched records as well. 

[BenMoss · Accepted solution]

The recommendation when you use the fuzzy match tool is to always follow it with a make group tool, the purpose of this is to identify all records which are linked in some way.

wts say record A has matched with record B and record B matches with Record c.

the make group tool will create a dataset indicating that these records are all linked together, and it will look something like...

A | A

A | B

A | C

You could then join your original source data to the make group tool and perform the aggregations you then desire.

This blog from Chris Love, though old, is a great video:

https://www.theinformationlab.co.uk/2014/04/02/alteryx-tools-focus-fuzzy-match-make-group-unique/

Ben