| Description |
1 online resource (1 streaming video file (31 min., 2 sec.)) |
| Performer |
Presenter, Kelsey Redman. |
| Note |
Title from resource description page (Safari, viewed October 29, 2020). |
|
Place of publication from title screen. |
| Summary |
"Presented by Kelsey Redman, AVP, Data Science at Comerica Bank. Purchasing 3rd party data on individuals can give great insights on customers, but first we have to know which individuals from that outside data source are actually customers and which are just prospects. Without a unique identifier like SSN or Driver's License number from the 3rd party data, we have to use a combination of name, address, and demographic information to identify the matching customer. Between nicknames, misspelled names and addresses, and family members with similar names all at one address, this quickly becomes a difficult task involving heavy data cleanup and an increasingly complicated series of rules. In this presentation, we demonstrate some techniques to help resolve these entities across data sources by employing the use of supervised classification machine learning techniques to quantify and predict entity 'likeness.' We showcase some of the challenges we faced with exploring other entity resolution methods, with manually labeling a comprehensive training set, and how this approach might extend to solve other data issues."--Resource description page |
| Subject |
Machine learning.
|
|
Consumers -- Identification.
|
|
Electronic data processing.
|
|
Apprentissage automatique. |
|
Consumers. |
|
Electronic data processing. |
|
Machine learning. |
| Genre |
Field guides.
|
| Added Author |
Redman, Kelsey, on-screen presenter.
|
|
Data Science Salon, publisher.
|
|
More Resources