Fuzzy logic for transposed letters in Address MatchBatch transform

A ‘Single Customer View’ or ‘Single Citizen View’ is the linking of customer data to provide an accurate and holistic view of any one customer across different channels and lines of business. One of the common ways to achieve this is to compare addresses of different customers and group them based on the common address.

The Address_MatchBatch Transform can be used to group together similar addresses. While comparing the address we want to compare “similar” addresses instead of going for “Exact” match. Each “Similar” address is given a score indicating the degree of Match.  An Exact match will yield a score of 100.

There is a specific option to cater to transposed letters like “ABCD” and “ABDC”.

If we select Check for transposed letters=Yes then the transform deducts half as many points for transposed characters as it deducts for other non-matching characters. This can be explained better with an example:

The input data has values “MATHURA” and MATHUAR” in the column mapped to Address Primary Name. The address Match score differs when we run the transform with Check for transposed letters=Yes and Check for transposed letters=No