Skip to contents
synpmx
0.0.0.9000
Reference
Background
The synpmx data generation algorithms
Comparing AVATAR, PCA and PMXmodel
Literature: generating synthetic data
Privacy
What differential privacy does and does not guarantee
Where d, f, and the error law come from
The mechanism-level privacy argument
Feasibility by cohort size
Elicitation, for the modes that assert a model
Building the structural model
Describing the trial
1. PMXmodel
Demo
Algorithm
Evaluation on public data
Study fingerprint
Privacy protections and checks
Membership inference, by simulation
2. AVATAR
Demo
Algorithm
Evaluation on public data
Example: are relationships preserved?
Addl. Methods
3. PCA
Demo
Algorithm
Evaluation on public data
Study fingerprint
4. Prior
Demo
Algorithm
Evaluation on public data
5. Calibrated
Demo
Algorithm
Evaluation on public data
Scorecard
Checks of the synthetic data
Literature: checking synthetic data
License
YEAR: 2026 COPYRIGHT HOLDER: Andrew Stein