Attacking primary suppressed cells

Function [attack()] is used to compute lower and upper bounds for a given sdcProblem instance. For all calculations the current suppression pattern is used when calculating solutions of the attacker's problem.

Usage

attack(
  object,
  to_attack = NULL,
  verbose = TRUE,
  threshold = 1e-08,
  n_workers = 1,
  ...
)

Arguments

object: an object of class `sdcProblem`
to_attack: if `NULL` all current primary suppressed cells are attacked; otherwise either an integerish (indices) or character-vector (str-ids) of the cells that should be attacked.
verbose: a logical scalar determining if additional output should be displayed
threshold: a numeric scalar defining the tolerance for the protection check. A cell is considered protected if the absolute difference between the computed upper and lower bound is strictly greater than this threshold. Defaults to `1e-8`.
n_workers: a scalar positive integer specifying the number of parallel workers to use for calculation
...: placeholder for possible additional input, currently unused;

Value

a `data.frame` with the following columns: - `sdc`: the original sdc-status code - `freq`: the original value of the cell - `id`: index of primary suppressed cells - `low`: computed lower bound of the attacker's problem - `up`: computed upper bound of the attacker's problem - `protected`: logical, TRUE if the absolute difference between `up` and `low` exceeds the defined `threshold`

Author

Bernhard Meindl bernhard.meindl@statistik.gv.at

Examples

if (FALSE) { # \dontrun{
dims <- list(
  v1 = sdcHierarchies::hier_create("tot", letters[1:4]),
  v2 = sdcHierarchies::hier_create("tot", letters[5:8])
)

N <- 150
df <- data.frame(
  v1 = sample(letters[1:4], N, replace = TRUE),
  v2 = sample(letters[5:8], N, replace = TRUE)
)

sdc <- makeProblem(data = df, dimList = dims)

# set primary suppressions
specs <- data.frame(
  v1 = c("a", "b", "a"),
  v2 = c("e", "e", "f")
)
sdc <- change_cellstatus(sdc, specs = specs, rule = "u")

# attack all primary sensitive cells with a custom threshold
attack(sdc, to_attack = NULL, threshold = 1e-6)

# protect the table and attack again
sdc <- protectTable(sdc, method = "SIMPLEHEURISTIC")
attack(sdc, to_attack = NULL)

# attack only selected cells
attack(sdc, to_attack = c(7, 12))

# Parallel processing (future.apply needs to be available)
attack(sdc, n_workers = 4)
} # }