Utility Analysis

Data utility means the usefulness of the anonymised data for statistical analyses by data analysts as well as the validity of these analyses when performed on the anonymised data.

Equivalence class refers to the group of indistinguishable records sharing the same values on a set of indirect identifiers. Refer to 𝑘-anonymity for a better understanding.

Utility Metrics in Cloak

In Cloak, the following utility metrics are computed after applying 𝑘-anonymity:

• Maximum equivalence class size
  The maximum size of the equivalence classes.

• Average equivalence class size
  The average size of the equivalence classes.

• Information loss
  The loss of utility due to a reduction in accuracy (with generalisation) or complete loss (with suppression) of indirect identifiers’ values.

• Normalised average equivalence class size
  Measures how well the creation of the equivalence classes approaches the best case, where each record is generalised in an equivalence class of k records.
  The objective is to minimise the penalty: a value of 1 would indicate the ideal anonymisation in which the size of the equivalence classes equals the given k value.

Need help? Try our Support Assistant or visit the Support page.