NEWS.md
ck_setup()
ptable
CK_RUN_PARALLEL. If this is set to TRUE, parallel computation is enabled, otherwise it is disabled. By default, parallel computing is disabled.ptable input object$supp_cells() that allows to specifiy sensitive cells based on namessign() with ifelse to enforce perturbation of cells that require additional protectionw = NULL in ck_setup()
digest-pkgroxygen2
hierarchy_info() containing some important information for each dimensioncreate_cnt_ptable() and create_num_ptable() from ptable-pkg
ck_params_nums()
ptable::pt_create_pParams as input in ck_params_cnts()
ptable::pt_create_pParams as input in ck_params_nums()
issue #131
ptable-pkgptable-pkg using ptable::pt_ex_cnts() and ptable::pt_ex_nums()
freqtab()
numtab()
params_cnts_set() and params_nums_set()
params_cnts_get() and params_nums_get()
allvars(), numvars() and cntvars() returning variable names eligable for perturbationck_params_nums() to define perturbation parameters for continuous variables along with helper-functions ck_flexparams() and ck_simpleparams()
numtab() to extract numerical tablesmod_nums() returning modifications for numerical variablesprint() and summary() to include information about perturbed continuous variablesreset_cntvars(), reset_numvars() and reset_allvars() to remove perturbation results and provided perturbation parameters$supp_freq(v, max_n)$supp_nk(v, max_n)$supp_p(v, max_n)$supp_pq(v, max_n)ptable-pkgck_params_nums() and ck_params_cnts()
ck_read_yaml()
exclude_zero_cells to ck_cnt_measures()
sdcHierarchies to define hierarchiesptable-pkg is not yet definedABS definitionby-argument in $perturb() methodR6 classes
ck_setup() to define a table$perturb()-methodck_export_table() and added arguments to method freqtab()
$print() for R6 objects$summary() for R6 objects$mod_cnts() returning modifications for count variables$params_cnts() that allow to query and set count parametersck_cnt_measures()
false_positives to false_nonzero
ck_cnt_measures showing exact perturbations$measures() uses ck_cnt_measures() internally for count variablessdcHierarchies to generate and update hierarchiesck_rename_nodes()
perturbTable() got a new argument pThreshold that allows to specify a threshold above no perturbation is applied independent from the perturbation table. Currently only a placeholder and not used.ck_vignette() that displays the package vignette in a browserck_generate_rkeys() got a new argument seed that allows to overwrite the default seed computed from a hash of the input dataset.ck_export_table() that allows to save results in a simple formatbigN are specifiedsmallN is too large in respect to the specified pTableck_generate_rkeys() only if non-required parameters have been actually specifiedck_cnt_measures_basic() that computes infoloss/utility measures based on two input vectors referring to original and perturbed valuesck_cnt_measures() that computes some (distance-based) information loss measures for count variablesck_create_pTable(); thus the package must be installed, e.g using devtools::install_github("sdcTools/ptable", build_vignette=FALSE)
sdcTable and bump version requirement of this package to >=0.23
by in perturbTable() it is automatically added to countVars even though not explicitely specified.perturbTable() gained an optional new argument by. In this argument one can use a variable that must also be listed in countVars. This variable is then used to compute the magnitute tables by the given 0/1 binary variable. For an example see ?perturbTable.countVars in perturbTable() which allows to additionally tabulate any number or 0/1 variables. For such variables. In such case, the record-keys of non-contribution units are set to 0 prior to the lookup in the perturbation tableresults() and replaced it with new methods ck_freq_table() and ck_cont_table() that should be used to query specific tables from the output of perturbTable()
perturbTable(), ck_freq_table() and ck_cont_table()
feature: new dynamic way to specify hierarchies for tables, for an example see ?ck_manage_hierarchies. This functionality will eventually also find its way to sdcTable
pTable) can now be specified in two different formats. The (default) way is to specify it as described in the original ABS-paper Methodology for the Automatic Confidentialisation of Statistical Outputs from Remote Servers at the Australian Bureau of Statistics (Thompson, Broadfoot, Elazar). An alternative way is to provide the perturbation tables for count tables in the “destatis”-format. ck_create_pTable(type="destatis") returns an exemplary pTable in this format. In the future, such pTables will likely be generated from another package. As the requirements regarding record keys are different in the following lookup-approach, we have already implemented some (basic) checks for validity of record keys when they are already available in the microdata used in ck_create_input().