AttributeInferenceAttackCAP#

synthesized.insight.metrics.privacy.AttributeInferenceAttackCAP

class AttributeInferenceAttackCAP(model, sensitive_col, predictors=None)#

Computes the privacy score between two dataframes using CAP (Correct Attribution Probability) method

Parameters
  • model (str) – ‘GeneralizedCAP’ or ‘DistanceCAP’

  • sensitive_col (str) – Name of the sensitive column

  • predictors (Optional[List[str]]) – List of predictor column names. If not provided, then all columns except the sensitive column is treated as predictors

Returns

Privacy score between 0 and 1

Methods

Attributes

name

tags