|
Activity Number:
|
465
- Probabilistic Record Linkage: Better Assumptions, Scalable Inference, and Accounting for Uncertainty
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Social Statistics Section
|
|
Abstract #329092
|
|
|
Title:
|
Bayesian Record Linkage with Sub-Models
|
|
Author(s):
|
Joan Heck*
|
|
Companies:
|
|
|
Keywords:
|
Bayesian;
Record Linkage;
Voting
|
|
Abstract:
|
Record linkage is used to merge data sets in the absence of unique identifiers, relying on comparisons of shared fields to do so. Most common algorithms assume independence of these field comparisons conditional on whether or not they correspond to the same entity. We propose a Bayesian approach to record linkage with a loosened assumption of conditional independence, allowing the conditional distribution of comparisons to change based on the values of other fields. We apply our model to North Carolina 2016 voting data, linking records of provisional voters to the 2016 NC voter file.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2018 program
|
