recordSwapping

Package recordSwapping has been moved to sdcMicro! Please use package sdcMicro and see ?sdcMicro::recordSwap() or vignette('recrodSwapping') for details and examples.\n CRAN: https://github.com/sdcTools/sdcMicro/packages/sdcMicro/index.html\n Development: https://github.com/sdcTools/sdcMicro

recordSwapping was an R-package for record swapping. Its aim was to develop a core library using pure C++ (11) code to implement targeted record swapping (TRS). Since version 1.0.1, the functionality has been included in sdcMicro (versions >= 5.7.1) where also any future development will take place.

The current files of the "core" library can be found in github.com/sdcTools/sdcMicro/tree/master/src/recordSwap. This folder contains the pure C++ (11) code recordSwap.cpp and recordSwap.h which should be used if the functionality should be included in other software tools. The C++ code is not directly embedded using Rcpp so that it can be more easily implemented into other projects which do not depend on R libraries.

Within sdcMicro, the library is called by an Rcpp wrapper (src/recordSwap_R.cpp) which again is called from the top level R-function recordSwap().

Versions and Updates

Below are the changes between versions shown; Versions > 1.0.1 will be documented in sdcMicro.

1.0.1

• fix bug when producing the log file.
• If a household has no suitable donor household once, then this household is discarded from swapping for all subsequent swaps in all other geographic hierarchies. Otherwise households which are at-risk and have no suitable donor at a specific geographic hierarchy can be swapped in lower geographic hierarchies. This leads to the household being swapped but the swap does not account for the at-risk issue.
• Improved inputs checks for recordSwap(), thanks @Kyoshido for PR 67093f7
• Corrected spelling on vignette and documentation, thanks @Kyoshido for PR 4dfb068

1.0.0

• included parameters int &count_swapped_records and int &count_swapped_hid std::string log_file_name to the cpp-function recordSwap() which count the number of swapped records and swapped households. These parameters are convenience parameters for mu-Argus.
• included parameter std::string log_file_name and log_file_name in the cpp-function recordSwap() and R-function recordSwap() respectively. Contains path for writing a log file. The log file contains a list of household IDs (hid) which could not have been swapped and is only created if any such households exist.
• Changed definition of parameter k_anonymity: now a household is treated as high-risk-household if for at least 1 person in the household counts < k_anonymity, was previously counts <= k_anonymity. This definition is now consistent with parameter k of function sdcMicro::localSuppression().

0.4.0

• Made recordSwap() an S3 Method, which can also accept an sdcMicro-object as input.
• included unit tests for recordSwap() when using sdcMicro-objects
• included unit tests for infoLoss()
• cleaned up code and unit tests such that R CMD check runs without notes, warnings and errors
• fixed warnings when compiling c++ code like comparing unsigned integer ... or unused variable ...
• fixed minor typos in README, help pages and vignettes

0.3.3

• included function infoLoss() to calculated various information loss measures after recordSwap() was applied. Function infoLoss() is heavily inspired by function cellKey::ck_cnt_measures() but accepts the micro data as input instead of frequency or magnitude counts.
• changed variable names and content of dummy data created by createDat.
• updated documentation and fixed some typos.

0.3.2

• extended unit tests
• Argument data for R-function recordSwap() does no longer need to have only integer values. Only variables needed for underlying C++ function defined through parameter hid, hierarchy, risk_variables, similar, carry_along need to be integer type.

0.3.1

• Fixed bug where less households than indicated through the swaprate were drawn
• added version number to header file
• developed different method for distributing number of draws over all geographic hierarchies

0.3.0

• Changed parameter order in R and C++ function recordSwap() to have a more consistent documentation. Function call to C++ function changed to

std::vector< std::vector<int> > recordSwap(std::vector< std::vector<int> > data, int hid,
                                           std::vector<int> hierarchy, 
                                           std::vector< std::vector<int> > similar,
                                           double swaprate,
                                           std::vector< std::vector<double> > risk, double risk_threshold,
                                           int k_anonymity, std::vector<int> risk_variables,  
                                           std::vector<int> carry_along,
                                           int seed = 123456)

• Improved interface of R function recordSwap(). Now, parameters hid, hierarchy, similar and risk_variables can be used with column indices or column names of data. Please note that indices in R start with 1 but in C++ they start with 0. The R function recordSwap() which is basically just a wrapper for the C++ function converts column names or indices into the correct format for the C++ function. So the call for recordSwap() from R expects indices starting from 1.

• added parameter std::vector<int> carry_along to C++ function recordSwap(). Like the paramter hierarchy, carry_along expects column indices of data and swaps these values in addition to the ones defined in hierarchy. These variables do however not interfere with risk calculation, sampling or finding a donor.

• added parameter carry_along to R function recordSwap(), expects either column indices or column names

• added parameter return_swapped_id to R function recordSwap(). If return_swapped_id is TRUE an additional column will be returned which holds the hid with which each record was swapped with. If this new column has the same value as hid the record was not swapped.

• Improved documentation and vignette

0.2.0

• Some parameter changes to the C++ function recordSwap():

  • similar is now a vector of vectors allowing for multiple similarity profiles
  • changed std::vector<int> risk to std::vector<int> risk_variables for a more descriptive name
  • changed int th to int k_anonymity for a more descriptive name
  • added parameter double risk_threshold which can be used to set a custom risk threshold. household whith a risk greater or equal risk_threshold is set as risky household (not yet supported by R wrapper)
  • added parameter std::vector<std::vector<double>> risk which can be used as custom risks for each household in each hierarchy level (not yet supported by R wrapper)

recordSwap(std::vector< std::vector<int> > data, std::vector<std::vector<int>> similar,
                                           std::vector<int> hierarchy, std::vector<int> risk_variables, int hid, 
                                           int k_anonymity, double swaprate, double risk_threshold,
                                           std::vector<std::vector<double>> risk, int seed = 123456)

• Fixed some bugs concerning the application of swaprate
• data as well as risk are used internally such that data[0] or risk[0] contain the micro data or risk over all hierarchies for the first individual, data[1] or risk[1] contain the micro data or risk, over all hierarchies, for the second individual, and so on.
• Documentation and parameter descriptions have been updated in the corresponding help files.

0.1.0

• First prototype version of record swapping
• Contains the function recordSwap() as the main function of this package
• recordSwap() is an R-wrapper which calls an underlying Rcpp function where at the bottom a call to the C++ function recordSwap() is made.

recordSwap(std::vector< std::vector<int>> data, std::vector<int> similar,
                                           std::vector<int> hierarchy, std::vector<int> risk, int hid, 
                                           int th, double swaprate, int seed = 123456)

• The parameter data contains the household data and is used internally such that data[0] contains values for each individual for the first column of the data, data[1] contains the values for each individual for the second column of the data. This also implies that data[0][0] addresses the first value of the first column and so on.
• The procedure expects the data to be ordered by household ID (column hid). Ordering inside each household is irrelevant.
• Various internal help functions are included. These are however not intended for flexible use and are only called from withing recordSwap()

Installation and Application

The functionality can be used by installing sdcMicro and using recordSwap()

install.packages("sdcMicro")

The application of TRS using the record-swapping library is shown in a package vignette

vignette("recordSwapping", package = "sdcMicro")

which is also available online. Further information is available in the help-pages (?sdcMicro::recordSwap)