Machine Learning Solution for Client’s Address Entity Resolution

Machine Learning Solution for Client's Address Entity Resolution

About The Client

US based small IT business that offers software platform data insights

Challenge

The batch system of the client had a low accuracy rate in matching new data from a customer with their internal data. Due to an outdated system in place for string matching, many address were being incorrectly matched or not matched at all. The client hope to find a new strategy to be implemented to increase matching accuracy.

Solution

Implemented a full 3 step process using the tools Apache Solr, Dedupe, and machine learning models to achieve accuracy of 84% in address matching between datasets

Standardized process and cleaned all incoming data before being matched with implemented solution

Results

0 %+

Increased accuracy of address matching by 30%+

0 %

Accuracy in address matching between datasets