Fuzzy Match for Names

[alkafalhas]

Hi

I have problem with the fuzzy match tools. I'm trying to use fuzzy match for names, examples as below:

A	B
ASMAYATI ABDULAZIM	ASMAYATI BINTI ABDULAZIM
AZIZANG BIN IBRAHIM	AZIZANG IBRAHIM
CANDAY LOH FOONG MANG	LOH FOONG MANG

However current settings, pick up wrongly for example like below:

Current setting Match threshold 90% with match style: Name

C	D
CHIAH YUN CHING	CHIA SHYAN CHING
LOKE HENG FATTY	LEE CHONG FATTY
NURA ALIYANA BINTI MOHAMMADAD	NURA LIYANA BINTI MUHAMADAD
WUN PEI KENG	WUN PEI KANG

Is there a way to fix this?

[Maskell_Rascal]

Hi @alkafalhas 

Can you provide either your workflow or sample data?

[ArtApa]

Hi @alkafalhas - I don't believe this is achievable on the level that you described. The problem is that something that you describe as a "wrong" match is objectively a better match. I tried different Match Functions and "wrongs" always tend to have a better matching score:

I'd recommend to look at other features in your data set or on the process holistically to achieve a desired outcome. Less elegant idea would be to create a table of exceptions and use it to remove "wrongs" from the workflow before Fuzzy Matching.

[apathetichell]

have you considered parsing out the  name into components and then trying a fuzzy match? I think the way it searches it gives equal weight to variation at different point in the name - however, as you've pointed out differing sections of the name may be more important. In your case, you'd probably want a 100% match on "last name" and differing levels of fuzzy match on the other parts.