The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.
First CRAN release. The method is unchanged from 0.1.0 and follows the published paper (Jamme, 2027, doi:10.1007/978-3-032-37883-5_9). This release focuses on robustness and documentation. The API may still evolve until 1.0.0.
pm_plot_risk_max() gains x_axis = "n" and
overlays the frontier found by pm_suggest_n()
(mark_frontier).pm_calib_dominance() now passes level to
the CI loss metric (it was always computed at 95%).sigma_nu,
n, beta or rho is no longer
affected by floating-point representation of seq()
grids.purrr and dplyr are no longer a
dependency.First working version of the package. The API is still unstable: function names and arguments may change without deprecation until version 1.0.0.
pm_params() gathers every choice in a single object:
the policy, organised by disclosure scenario (dominance,
prule, diff), and the three mechanism
parameters (sigma_nu, sigma_eps,
n), filled in as calibration proceeds.pm_calib_diff() returns the theoretical grid of
sigma_eps values meeting a given (beta, tau)
couple, with the loss floor they imply.plot() on that grid draws the risk-utility
frontier.pm_commit_diff() records the decision and sets
sigma_eps.pm_calib_dominance() returns, over a grid of dominance
levels rho, the scenario-I risk and the information loss
for each candidate (sigma_nu, n).summary() reduces that grid to the worst-case risk and
the loss range of every parameter combination.pm_suggest_n() solves for the largest n
meeting the risk ceiling — the calibration rule of step 2 — with an
optional safety margin.pm_commit_dominance() records sigma_nu and
n, and reports the resulting trade-off.pm_plot_risk_profile(), pm_plot_risk_max()
and pm_plot_tradeoff() reproduce the calibration figures of
the paper. The first two accept any scenario; the third is specific to
the risk metric selected.pm_plot_impact() superposes the empirical and
theoretical losses depending on rho.pm_perturb() applies
Y' = Y (1 + rho^n nu + eps) to an aggregated table and
returns it with rho, the derived keys and the perturbed
total, plus a pm_meta attribute recording the full
configuration.pm_draws() derives the two Gaussian draws from a cell
key by SHA-512 hashing and quantile inversion, so that perturbations are
reproducible over time.assess_loss_expectation() and
assess_loss_ci() give the a-priori information loss
conditional on dominance.assess_risk_I() and assess_risk_II() give
the a-priori risk measures for the external and internal inference
scenarios.assess_risk_diff() gives an a-priori upper bound of the
risk measure for the differencing scenario.assess_utility_empirical() and
assess_risk_empirical() measure what was actually achieved
on the perturbed table, alongside the theoretical values predicted for
the same cells.These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.