Measure risk by estimating inverse transition probabilities

Calculates probabilities P(X=i|X'=j) where X denotes the original value and X' the perturbed value, providing risk measures for statistical disclosure control.

Usage

assess_risk(matrice_transition, freq, I, J)

Arguments

matrice_transition

ptable object. Object returned by create_transition_matrix

freq

data.frame. Object returned by compute_frequencies

I

integer vector. Original values to consider

J

integer vector. Perturbed values to consider

Value

data.frame with 5 columns:

• i: original value(s)

• j: perturbed value(s)

• pi_hat: estimated probability P(X = i)

• pij: transition probability P(X' = j | X = i)

• qij: inverse transition probability P(X = i | X' = j)

Examples

if (FALSE) { # \dontrun{
library(ptable)
library(dplyr)
mat_trans <- create_transition_matrix(D = 5, V = 2)
data("dtest")

tab_comptage <- tabulate_cnt_micro_data(
  df = dtest, rk = NULL,
  cat_vars = c("DEP", "DIPLOME", "SEXE", "AGE"),
  marge_label = "Total",
  freq_empiriq = TRUE
)

# Calculate inverse transition probabilities P(X=i|X'=1) with i in 1:4
assess_risk(mat_trans, tab_comptage$freq, 1:4, 1)

# Calculate for multiple i and j values
assess_risk(mat_trans, tab_comptage$freq, 1:4, 1:4)
} # }