Package index • dominoise

Parameters

The single object gathering every policy and mechanism choice, filled in as calibration proceeds.

pm_params(): Create the perturbation-parameter object containing all the parameters of the noise.

Calibration — step 1, differencing noise

Choosing sigma_eps from the differencing risk ceiling. This parameter also sets the loss floor borne by every cell.

pm_calib_diff(): Calibration of the differencing noise (sigma_epsilon)
pm_sigma_eps(): Smallest sigma_eps guaranteeing the differencing-risk ceiling
pm_commit_diff(): Commit the differencing decision into the parameter object

Calibration — step 2, dominance noise

Choosing sigma_nu and n from the scenario-I risk ceiling, keeping the largest admissible n.

pm_calib_dominance(): Calibration of the dominance noise (sigma_nu and n)
pm_suggest_n(): Largest admissible shape parameter n
pm_commit_dominance(): Commit the dominance decision (sigma_nu and n) into the parameter object

Calibration figures

The risk and risk-utility figures of the paper.

pm_plot_risk_max(): Worst-case risk as a function of sigma_nu
pm_plot_risk_profile(): Risk profile as a function of dominance
pm_plot_tradeoff(): Risk-utility trade-off map

Applying the mechanism

Deriving the Gaussian draws from the cell keys, and perturbing an aggregated table.

pm_draws(): Derive the two Gaussian draws from a cell key
pm_perturb(): Perturb an aggregated table

A-priori risk and utility

Closed-form metrics, conditional on the dominance level, computed from the parameters alone.

assess_loss_expectation(): Conditional expectation of the absolute relative loss |Z|
assess_loss_ci(): Confidence-interval bound of the relative loss Z
assess_risk_I(): Scenario-I (dominance) risk measure
assess_risk_II(): Scenario-II (p%-rule) risk measure
assess_risk_diff(): Compute the upper bound of the differencing risk, following the proposition 7 of the paper.

Ex-post assessment

Risk and utility measured on the perturbed table itself.

assess_utility_empirical(): Observed information loss
assess_risk_empirical(): Observed disclosure risk

Methods and utilities

S3 methods for printing, summarising and plotting the objects returned by the calibration functions. Not called directly.

summary(<pm_calib_dominance>): Worst-case risk and loss range of a dominance calibration grid
plot(<pm_calib_diff>): Plot method for a differencing calibration grid
plot(<pm_calib_dominance>): Plot method for a dominance calibration grid