Free Engineering Workbench calculator
Re-identification risk (k-anonymity)
Estimate re-identification risk of a shared dataset from its equivalence class sizes.
Privacy & data protectionHow it's worked out
- Maximum risk = 1 ÷ k. Average risk = classes ÷ records
- Quasi-identifiers chosen correctly; attacker knows the target is in the data
Notek-anonymity does not stop attribute disclosure; combine with l-diversity for sensitive values.
Source: Sweeney (2002); El Emam, Guide to the De-Identification of Personal Health Information; ISO/IEC 20889
Need help acting on this number?
VVnT SeQuor's Privacy & Data Protection practice helps teams turn this metric into a plan.