Understanding Why z_max is 0.75 Despite Strong Matches in Filtered z_cluster Data

·Apr 04, 2025 04:06 AM

Hey Good Morning When we filter for a specific z_cluster, I notice that the matches are very strong, with most of the entity columns being identical in each row. However, we're still getting a z_max of 0.75. Is there any particular reason for this?

21 comments

· Sorted by Oldest

Monish K.
·
Adding examples FYI
Monish K.
·
Also pls hare the confidence score calculation using these Z scores..!!
Monish K.
·
Response Pls
Sonal G.
·
Are these the only fields you run the match on ?
Sonal G.
·
Typically the scoring reflects the labelling
Sonal G.
·
If there are fields without information but part of matching, they will dip the score even if other fields match
Sonal G.
·
HTH
Monish K.
·
these are the only fields used fuzzy, please help with score calculation formula using these Zscores
Sonal G.
·
Can you share the field definition?
Monish K.
·
customer_contact_email = FieldDefinition("email", "string", MatchType.DONT_USE) mobile_number = FieldDefinition("phone_number", "string", MatchType.DONT_USE) customer_fullname = FieldDefinition("full_name", "string", MatchType.FUZZY) city = FieldDefinition("city", "string", MatchType.FUZZY) state = FieldDefinition("state", "string", MatchType.FUZZY) zipcode = FieldDefinition("zipcode", "string", MatchType.FUZZY) Vertical_name = FieldDefinition("Vertical_name", "string", MatchType.DONT_USE)
Sonal G.
·
Thank you. Are there other records in the cluster besides the ones you shared?
Monish K.
·
yes we do have it.
Sonal G.
·
ah. thats the reason. min score is the least it matched to another record int he cluster.
Sonal G.
·
max is the best it matched to another record in the cluster
Monish K.
·
Please provide the confidence score calculation formula using these Zscores