|
Activity Number:
|
212
- GOVT CSpeed 1
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 10, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Government Statistics Section
|
|
Abstract #317710
|
|
|
Title:
|
Implementing Differentially Private Geometric Mechanism to Business Formation Statistics and Manufactured Housing Survey for Privacy Protection
|
|
Author(s):
|
THAGENDRA PRASAD TIMSINA*
|
|
Companies:
|
CENSUS BUREAU
|
|
Keywords:
|
Sensitivity;
Privacy Loss Budget;
Differential Privacy;
Geometric Mechanism;
Relative Accuracy;
Post Processing
|
|
Abstract:
|
Abstract: The U.S. Census Bureau recently received data requests for annual tabulations of Business Formation Statistics (BFS): business applications by county and Manufactured Housing Survey(MHS): total shipment count estimates by Core-Based Statistical Area (CBSA) and size of home (Single-section, Multi-section). Both tabulations contain confidential information and the privacy of each record that feature in these tabulations are protected by law. We implemented a differentially private geometric mechanism for releasing these tabulations. This mechanism allows us to choose between amounts of privacy we are willing to give up and the accuracy of the data. We used l_1 error metric and relative accuracy to find an optimal tradeoff between privacy and accuracy.
|
Authors who are presenting talks have a * after their name.
