Fuzzy Matching
Fuzzy Matching is the reconciliation of “non-identical duplicates” between two or more data sources that couldn’t otherwise be joined together by more traditional methods.
Below illustrates an Alteryx use case Continuum completed successfully for an international bank which needed to show that data being screened in 2 systems was consistent.
Two outputs were required…
Matches between the two sources with level of confidence.
Exceptions from source A, ready to be migrated to source B
Data is queried, prepared and cleansed in Alteryx
Both datasets are then transformed into the same format, ready for matching
Attempt to match “100%” by a combination of [Name] and wide array of other KDEs (Key Data Elements) such as date of birth, address, passport number etc.
Records that do not match are passed on for Fuzzy Matching.
Records are passed through multiple “fuzzy matches” with varying match thresholds.
Each record then has a potential range from full match to no match, with a varying level of confidence in between.
The client is then able to make a risk-based decision on whether data matches between the systems.
Matched pairs are joined together and stored ready for reconciliation
Exceptions from source A are isolated ready for migration to source B
