Membership

The membership of the synthetic data project is now at 38.

Scope

The goal of the Synthetic Data Project is to develop a hands-on guide for creating and using synthetic data primarily geared towards data protection and disclosure control. The target audience of this guide includes both managers and methodologist at NSOs to provide a starting point for using and creating synthetic data. 

After the HLG-MOS Executive Board recommendations, the guide will emphasize creating synthetic data from non-survey data sources. In addition, the use cases will highlight the circumstances in which to synthetic data is the appropriate solution to their output privacy problems. 

Work Packages

Work package 2 - methodology - has been under with presentations of new method from Statistics New Zealand and Australian Bureau of Statistics. 

Patnerships

The HLG-MOS Executive Board recommended partnerships with the ML 2021 group and the Input Privacy Preserving Techniques project. I met with both projects and synthetic data can be a suitable option for testing data. 

Work Packages

Work Package 1 - Use Cases: the BSTN Synthetic Data Working Group in 2020 had created a foundation of use cases for synthetic data. After the EB's recommendations, the use case categories are being reassessed. The use case work package has two main outputs:

1. A description of different use cases and highlighting their analytical and disclosure risk needs
2. Creating a gradient of different use cases based on their analytical and disclosure risk needs

The use cases will form the basis for the methodology and quality measure recommendations. 

The target deadline for work package 1 is end of May 2021. 

Work Package 2 - methodology: The following methodology presentations have been scheduled:

• March 4, 2021: Joerg Drechsler presenting on method used to synthesize US business data
• April 1, 2021: Kenza Sallier presenting Statistics Canada’ experience creating public synthetic datasets using the FCS and the Synthpop package and Gillian Raab and Beata Nowok will be presenting on new additions to the Synthpop package. 
• May 6, 2021: Tentative Christine Task and Rolando Rodriguez

Patnerships

The ML 2021 group will get back to me on their requirements for testing data and we will then determined if synthetic data is the right solution.