Title: | Small Count Rounding of Tabular Data |
---|---|
Description: | A statistical disclosure control tool to protect frequency tables in cases where small values are sensitive. The function PLSrounding() performs small count rounding of necessary inner cells so that all small frequencies of cross-classifications to be published (publishable cells) are rounded. This is equivalent to changing micro data since frequencies of unique combinations are changed. Thus, additivity and consistency are guaranteed. The methodology is described in Langsrud and Heldal (2018) <https://www.researchgate.net/publication/327768398_An_Algorithm_for_Small_Count_Rounding_of_Tabular_Data>. |
Authors: | Øyvind Langsrud [aut, cre], Johan Heldal [aut] |
Maintainer: | Øyvind Langsrud <[email protected]> |
License: | MIT + file LICENSE |
Version: | 1.0.8 |
Built: | 2024-10-23 06:19:18 UTC |
Source: | https://github.com/statisticsnorway/ssb-smallcountrounding |
A statistical disclosure control tool to protect frequency tables in cases where small values are sensitive.
The main function, PLSrounding
, performs small count rounding of necessary inner cells (Heldal, 2017)
so that all small frequencies of cross-classifications to be published (publishable cells) are rounded.
This is equivalent to changing micro data since frequencies of unique combinations are changed.
Thus, additivity and consistency are guaranteed.
This is performed by an algorithm inspired by partial least squares regression (Langsrud and Heldal, 2018).
Maintainer: Øyvind Langsrud [email protected]
Authors:
Johan Heldal
Heldal, J. (2017): “The European Census Hub 2011 Hypercubes - Norwegian SDC Experiences”. In: Work Session on Statistical Data Confidentiality, Skopje, The former Yugoslav Republic of Macedonia, September 20-22 , 2017.
Langsrud, Ø. and Heldal, J. (2018): “An Algorithm for Small Count Rounding of Tabular Data”. Presented at: Privacy in statistical databases, Valencia, Spain. September 26-28, 2018. https://www.researchgate.net/publication/327768398_An_Algorithm_for_Small_Count_Rounding_of_Tabular_Data
Useful links:
FormulaSelection method for PLSrounded
## S3 method for class 'PLSrounded' FormulaSelection(x, formula = NULL, intercept = NA, logical = FALSE)
## S3 method for class 'PLSrounded' FormulaSelection(x, formula = NULL, intercept = NA, logical = FALSE)
x |
PLSrounded object |
formula |
|
intercept |
|
logical |
|
Limited version of the publish data frame
Hellinger distance (HD
) and a related utility measure (HDutility
)
described in the reference below.
The utility measure is made to be bounded between 0 and 1.
HD(f, g) HDutility(f, g)
HD(f, g) HDutility(f, g)
f |
Vector of original counts |
g |
Vector of perturbed counts |
HD is defined as "sqrt(sum((sqrt(f) - sqrt(g))^2)/2)
" and
HDutility is defined as "1 - HD(f, g)/sqrt(sum(f))
".
Hellinger distance or related utility measure
Shlomo, N., Antal, L., & Elliot, M. (2015). Measuring Disclosure Risk and Data Utility for Flexible Table Generators, Journal of Official Statistics, 31(2), 305-324. doi:10.1515/jos-2015-0019
f <- 1:6 g <- c(0, 3, 3, 3, 6, 6) print(c( HD = HD(f, g), HDutility = HDutility(f, g), maxdiff = max(abs(g - f)), meanAbsDiff = mean(abs(g - f)), rootMeanSquare = sqrt(mean((g - f)^2)) ))
f <- 1:6 g <- c(0, 3, 3, 3, 6, 6) print(c( HD = HD(f, g), HDutility = HDutility(f, g), maxdiff = max(abs(g - f)), meanAbsDiff = mean(abs(g - f)), rootMeanSquare = sqrt(mean((g - f)^2)) ))
Output from PLSrounding
is presented as two-way table(s) in cases where this is possible.
A requirement is that the number of main dimensional variables is two.
PLS2way(obj, variable = c("rounded", "original", "difference", "code"))
PLS2way(obj, variable = c("rounded", "original", "difference", "code"))
obj |
Output object from |
variable |
One of |
When parameter "variable"
is "code"
, output is coded as "#"
(publish), "."
(inner) and "&"
(both).
A data frame
# Making tables from PLSrounding examples z <- SmallCountData("e6") a <- PLSrounding(z, "freq", formula = ~eu * year + geo) PLS2way(a, "original") PLS2way(a, "difference") PLS2way(a, "code") PLS2way(PLSrounding(z, "freq", formula = ~eu * year + geo * year), "code") eHrc2 <- list(geo = c("EU", "@Portugal", "@Spain", "Iceland"), year = c("2018", "2019")) PLS2way(PLSrounding(z, "freq", hierarchies = eHrc2))
# Making tables from PLSrounding examples z <- SmallCountData("e6") a <- PLSrounding(z, "freq", formula = ~eu * year + geo) PLS2way(a, "original") PLS2way(a, "difference") PLS2way(a, "code") PLS2way(PLSrounding(z, "freq", formula = ~eu * year + geo * year), "code") eHrc2 <- list(geo = c("EU", "@Portugal", "@Spain", "Iceland"), year = c("2018", "2019")) PLS2way(PLSrounding(z, "freq", hierarchies = eHrc2))
Small count rounding of necessary inner cells are performed so that all small frequencies of cross-classifications to be published (publishable cells) are rounded. The publishable cells can be defined from a model formula, hierarchies or automatically from data.
PLSrounding( data, freqVar = NULL, roundBase = 3, hierarchies = NULL, formula = NULL, dimVar = NULL, maxRound = roundBase - 1, printInc = nrow(data) > 1000, output = NULL, extend0 = FALSE, preAggregate = is.null(freqVar), aggregatePackage = "base", aggregateNA = TRUE, aggregateBaseOrder = FALSE, rowGroupsPackage = aggregatePackage, ... ) PLSroundingInner(..., output = "inner") PLSroundingPublish(..., output = "publish")
PLSrounding( data, freqVar = NULL, roundBase = 3, hierarchies = NULL, formula = NULL, dimVar = NULL, maxRound = roundBase - 1, printInc = nrow(data) > 1000, output = NULL, extend0 = FALSE, preAggregate = is.null(freqVar), aggregatePackage = "base", aggregateNA = TRUE, aggregateBaseOrder = FALSE, rowGroupsPackage = aggregatePackage, ... ) PLSroundingInner(..., output = "inner") PLSroundingPublish(..., output = "publish")
data |
Input data as a data frame (inner cells) |
freqVar |
Variable holding counts (inner cells frequencies). When |
roundBase |
Rounding base |
hierarchies |
List of hierarchies |
formula |
Model formula defining publishable cells |
dimVar |
The main dimensional variables and additional aggregating variables. This parameter can be useful when hierarchies and formula are unspecified. |
maxRound |
Inner cells contributing to original publishable cells equal to or less than maxRound will be rounded |
printInc |
Printing iteration information to console when TRUE |
output |
Possible non-NULL values are |
extend0 |
When |
preAggregate |
When |
aggregatePackage |
Package used to preAggregate.
Parameter |
aggregateNA |
Whether to include NAs in the grouping variables while preAggregate.
Parameter |
aggregateBaseOrder |
Parameter |
rowGroupsPackage |
Parameter |
... |
Further parameters sent to |
This function is a user-friendly wrapper for RoundViaDummy
with data frame output and with computed summary of the results.
See RoundViaDummy
for more details.
Output is a four-element list with class attribute "PLSrounded",
which ensures informative printing and enables the use of FormulaSelection
on this object.
inner |
Data frame corresponding to input data with the main dimensional variables and with cell frequencies (original, rounded, difference). |
publish |
Data frame of publishable data with the main dimensional variables and with cell frequencies (original, rounded, difference). |
metrics |
A named character vector of various statistics calculated from the two output data frames
(" |
freqTable |
Matrix of frequencies of cell frequencies and absolute differences.
For example, row " |
Langsrud, Ø. and Heldal, J. (2018): “An Algorithm for Small Count Rounding of Tabular Data”. Presented at: Privacy in statistical databases, Valencia, Spain. September 26-28, 2018. https://www.researchgate.net/publication/327768398_An_Algorithm_for_Small_Count_Rounding_of_Tabular_Data
RoundViaDummy
, PLS2way
, ModelMatrix
# Small example data set z <- SmallCountData("e6") print(z) # Publishable cells by formula interface a <- PLSrounding(z, "freq", roundBase = 5, formula = ~geo + eu + year) print(a) print(a$inner) print(a$publish) print(a$metrics) print(a$freqTable) # Using FormulaSelection() FormulaSelection(a$publish, ~eu + year) FormulaSelection(a, ~eu + year) # same as above FormulaSelection(a) # just a$publish # Recalculation of maxdiff, HDutility, meanAbsDiff and rootMeanSquare max(abs(a$publish[, "difference"])) HDutility(a$publish[, "original"], a$publish[, "rounded"]) mean(abs(a$publish[, "difference"])) sqrt(mean((a$publish[, "difference"])^2)) # Six lines below produce equivalent results # Ordering of rows can be different PLSrounding(z, "freq") # All variables except "freq" as dimVar PLSrounding(z, "freq", dimVar = c("geo", "eu", "year")) PLSrounding(z, "freq", formula = ~eu * year + geo * year) PLSrounding(z[, -2], "freq", hierarchies = SmallCountData("eHrc")) PLSrounding(z[, -2], "freq", hierarchies = SmallCountData("eDimList")) PLSrounding(z[, -2], "freq", hierarchies = SmallCountData("eDimList"), formula = ~geo * year) # Define publishable cells differently by making use of formula interface PLSrounding(z, "freq", formula = ~eu * year + geo) # Define publishable cells differently by making use of hierarchy interface eHrc2 <- list(geo = c("EU", "@Portugal", "@Spain", "Iceland"), year = c("2018", "2019")) PLSrounding(z, "freq", hierarchies = eHrc2) # Also possible to combine hierarchies and formula PLSrounding(z, "freq", hierarchies = SmallCountData("eDimList"), formula = ~geo + year) # Single data frame output PLSroundingInner(z, "freq", roundBase = 5, formula = ~geo + eu + year) PLSroundingPublish(z, roundBase = 5, formula = ~geo + eu + year) # Microdata input PLSroundingInner(rbind(z, z), roundBase = 5, formula = ~geo + eu + year) # Zero perturbed due to both extend0 = TRUE and zeroCandidates = TRUE set.seed(12345) PLSroundingInner(z[sample.int(5, 12, replace = TRUE), 1:3], formula = ~geo + eu + year, roundBase = 5, extend0 = TRUE, zeroCandidates = TRUE, printInc = TRUE) # Parameter avoidHierarchical (see RoundViaDummy and ModelMatrix) PLSroundingPublish(z, roundBase = 5, formula = ~geo + eu + year, avoidHierarchical = TRUE) # Package sdcHierarchies can be used to create hierarchies. # The small example code below works if this package is available. if (require(sdcHierarchies)) { z2 <- cbind(geo = c("11", "21", "22"), z[, 3:4], stringsAsFactors = FALSE) h2 <- list( geo = hier_compute(inp = unique(z2$geo), dim_spec = c(1, 1), root = "Tot", as = "df"), year = hier_convert(hier_create(root = "Total", nodes = c("2018", "2019")), as = "df")) PLSrounding(z2, "freq", hierarchies = h2) } # Use PLS2way to produce tables as in Langsrud and Heldal (2018) and to demonstrate # parameters maxRound, zeroCandidates and identifyNew (see RoundViaDummy). # Parameter rndSeed used to ensure same output as in reference. exPSD <- SmallCountData("exPSD") a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, rndSeed=124) PLS2way(a, "original") # Table 1 PLS2way(a) # Table 2 a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, identifyNew = FALSE, rndSeed=124) PLS2way(a) # Table 3 a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, maxRound = 7) PLS2way(a) # Values in col1 rounded a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, zeroCandidates = TRUE) PLS2way(a) # (row3, col4): original is 0 and rounded is 5
# Small example data set z <- SmallCountData("e6") print(z) # Publishable cells by formula interface a <- PLSrounding(z, "freq", roundBase = 5, formula = ~geo + eu + year) print(a) print(a$inner) print(a$publish) print(a$metrics) print(a$freqTable) # Using FormulaSelection() FormulaSelection(a$publish, ~eu + year) FormulaSelection(a, ~eu + year) # same as above FormulaSelection(a) # just a$publish # Recalculation of maxdiff, HDutility, meanAbsDiff and rootMeanSquare max(abs(a$publish[, "difference"])) HDutility(a$publish[, "original"], a$publish[, "rounded"]) mean(abs(a$publish[, "difference"])) sqrt(mean((a$publish[, "difference"])^2)) # Six lines below produce equivalent results # Ordering of rows can be different PLSrounding(z, "freq") # All variables except "freq" as dimVar PLSrounding(z, "freq", dimVar = c("geo", "eu", "year")) PLSrounding(z, "freq", formula = ~eu * year + geo * year) PLSrounding(z[, -2], "freq", hierarchies = SmallCountData("eHrc")) PLSrounding(z[, -2], "freq", hierarchies = SmallCountData("eDimList")) PLSrounding(z[, -2], "freq", hierarchies = SmallCountData("eDimList"), formula = ~geo * year) # Define publishable cells differently by making use of formula interface PLSrounding(z, "freq", formula = ~eu * year + geo) # Define publishable cells differently by making use of hierarchy interface eHrc2 <- list(geo = c("EU", "@Portugal", "@Spain", "Iceland"), year = c("2018", "2019")) PLSrounding(z, "freq", hierarchies = eHrc2) # Also possible to combine hierarchies and formula PLSrounding(z, "freq", hierarchies = SmallCountData("eDimList"), formula = ~geo + year) # Single data frame output PLSroundingInner(z, "freq", roundBase = 5, formula = ~geo + eu + year) PLSroundingPublish(z, roundBase = 5, formula = ~geo + eu + year) # Microdata input PLSroundingInner(rbind(z, z), roundBase = 5, formula = ~geo + eu + year) # Zero perturbed due to both extend0 = TRUE and zeroCandidates = TRUE set.seed(12345) PLSroundingInner(z[sample.int(5, 12, replace = TRUE), 1:3], formula = ~geo + eu + year, roundBase = 5, extend0 = TRUE, zeroCandidates = TRUE, printInc = TRUE) # Parameter avoidHierarchical (see RoundViaDummy and ModelMatrix) PLSroundingPublish(z, roundBase = 5, formula = ~geo + eu + year, avoidHierarchical = TRUE) # Package sdcHierarchies can be used to create hierarchies. # The small example code below works if this package is available. if (require(sdcHierarchies)) { z2 <- cbind(geo = c("11", "21", "22"), z[, 3:4], stringsAsFactors = FALSE) h2 <- list( geo = hier_compute(inp = unique(z2$geo), dim_spec = c(1, 1), root = "Tot", as = "df"), year = hier_convert(hier_create(root = "Total", nodes = c("2018", "2019")), as = "df")) PLSrounding(z2, "freq", hierarchies = h2) } # Use PLS2way to produce tables as in Langsrud and Heldal (2018) and to demonstrate # parameters maxRound, zeroCandidates and identifyNew (see RoundViaDummy). # Parameter rndSeed used to ensure same output as in reference. exPSD <- SmallCountData("exPSD") a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, rndSeed=124) PLS2way(a, "original") # Table 1 PLS2way(a) # Table 2 a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, identifyNew = FALSE, rndSeed=124) PLS2way(a) # Table 3 a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, maxRound = 7) PLS2way(a) # Values in col1 rounded a <- PLSrounding(exPSD, "freq", 5, formula = ~rows + cols, zeroCandidates = TRUE) PLS2way(a) # (row3, col4): original is 0 and rounded is 5
The counts rounded by PLSrounding
Thereafter, based on the publishable rounded data, expected inner cell frequencies are generated by iterative proportional fitting using Mipf
.
To ensure that empty cells missing in input data are included in the fitting process, the data is first extended using Extend0
.
PLSroundingFits( data, freqVar = NULL, roundBase = 3, hierarchies = NULL, formula = NULL, dimVar = NULL, preAggregate = is.null(freqVar), printInc = nrow(data) > 1000, xReturn = FALSE, extend0 = FALSE, extend0Fits = TRUE, limit = 1e-10, viaQR = FALSE, iter = 1000, eps = 0.01, tol = 1e-13, reduceBy0 = TRUE, reduceByColSums = TRUE, reduceByLeverage = FALSE, ... )
PLSroundingFits( data, freqVar = NULL, roundBase = 3, hierarchies = NULL, formula = NULL, dimVar = NULL, preAggregate = is.null(freqVar), printInc = nrow(data) > 1000, xReturn = FALSE, extend0 = FALSE, extend0Fits = TRUE, limit = 1e-10, viaQR = FALSE, iter = 1000, eps = 0.01, tol = 1e-13, reduceBy0 = TRUE, reduceByColSums = TRUE, reduceByLeverage = FALSE, ... )
data |
data frame (inner cells) |
freqVar |
Variable holding counts |
roundBase |
Rounding base |
hierarchies |
List of hierarchies |
formula |
Model formula |
dimVar |
Dimensional variables |
preAggregate |
Aggregation |
printInc |
Printing iteration information |
xReturn |
Dummy matrix in output when |
extend0 |
|
extend0Fits |
When |
limit |
|
viaQR |
|
iter |
|
eps |
|
tol |
|
reduceBy0 |
|
reduceByColSums |
|
reduceByLeverage |
|
... |
Further parameters to |
The nine first parameters is documented in more detail in PLSrounding
.
If iterative proportional fitting succeeds, the maximum difference between rounded counts and ipFit
is less than input parameter eps
.
Output from PLSrounding
(class attribute "PLSrounded") with modified versions of inner
and publish
:
inner |
Extended with more input data variables and with expected frequencies ( |
publish |
Extended with aggregated expected frequencies ( |
z <- data.frame(geo = c("Iceland", "Portugal", "Spain"), eu = c("nonEU", "EU", "EU"), year = rep(c("2018","2019"), each = 3), freq = c(2,3,7,1,5,6), stringsAsFactors = FALSE) z4 <- z[-c(1:2), ] PLSroundingFits(z4, "freq", formula = ~eu * year + geo, extend0 = FALSE)[c("inner", "publish")] PLSroundingFits(z4, "freq", formula = ~eu * year + geo)[c("inner", "publish")] my_km2 <- SSBtools::SSBtoolsData("my_km2") # Default automatic extension (extend0Fits = TRUE) PLSroundingFits(my_km2, "freq", formula = ~(Sex + Age) * Municipality * Square1000m + Square250m)[c("inner", "publish")] # Manual specification to avoid Nittedal combined with another_km PLSroundingFits(my_km2, "freq", formula = ~(Sex + Age) * Municipality * Square1000m + Square250m, extend0Fits = list(c("Sex", "Age"), c("Municipality", "Square1000m", "Square250m")))[c("inner", "publish")] # Example with both extend0 (specified) and extend0Fits (default is TRUE) PLSroundingFits(my_km2, "freq", formula = ~(Sex + Age) * Municipality * Square1000m + Square250m, printInc = TRUE, zeroCandidates = TRUE, roundBase = 5, extend0 = list(c("Sex", "Age"), c("Municipality", "Square1000m", "Square250m")))[c("inner", "publish")]
z <- data.frame(geo = c("Iceland", "Portugal", "Spain"), eu = c("nonEU", "EU", "EU"), year = rep(c("2018","2019"), each = 3), freq = c(2,3,7,1,5,6), stringsAsFactors = FALSE) z4 <- z[-c(1:2), ] PLSroundingFits(z4, "freq", formula = ~eu * year + geo, extend0 = FALSE)[c("inner", "publish")] PLSroundingFits(z4, "freq", formula = ~eu * year + geo)[c("inner", "publish")] my_km2 <- SSBtools::SSBtoolsData("my_km2") # Default automatic extension (extend0Fits = TRUE) PLSroundingFits(my_km2, "freq", formula = ~(Sex + Age) * Municipality * Square1000m + Square250m)[c("inner", "publish")] # Manual specification to avoid Nittedal combined with another_km PLSroundingFits(my_km2, "freq", formula = ~(Sex + Age) * Municipality * Square1000m + Square250m, extend0Fits = list(c("Sex", "Age"), c("Municipality", "Square1000m", "Square250m")))[c("inner", "publish")] # Example with both extend0 (specified) and extend0Fits (default is TRUE) PLSroundingFits(my_km2, "freq", formula = ~(Sex + Age) * Municipality * Square1000m + Square250m, printInc = TRUE, zeroCandidates = TRUE, roundBase = 5, extend0 = list(c("Sex", "Age"), c("Municipality", "Square1000m", "Square250m")))[c("inner", "publish")]
The PLSrounding
runs are coordinated by using preliminary differences as input for the next run (parameter preDifference
)
PLSroundingLoop( data, loopId, ..., zeroCandidates = FALSE, forceInner = FALSE, preRounded = NULL, plsWeights = NULL, printInc = TRUE, preDifference = TRUE, preOutput = NULL, rndSeed = 123 )
PLSroundingLoop( data, loopId, ..., zeroCandidates = FALSE, forceInner = FALSE, preRounded = NULL, plsWeights = NULL, printInc = TRUE, preDifference = TRUE, preOutput = NULL, rndSeed = 123 )
data |
Input data as a data frame (inner cells) |
loopId |
Variable holding id for loops |
... |
|
zeroCandidates |
|
forceInner |
|
preRounded |
|
plsWeights |
|
printInc |
Printing iteration information to console when TRUE |
preDifference |
When TRUE, the |
preOutput |
preOutput The function can continue from output from a previous run |
rndSeed |
If non-NULL, a random generator seed to be set locally at the beginning of |
Note that in this function zeroCandidates
, forceInner
, preRounded
and plsWeights
cannot be supplied as vectors.
They may be specified as functions or as variables in the input data.
As output from PLSrounding
mf2 <- ~region + fylke * hovedint z2 <- SmallCountData("z2") a <- PLSroundingLoop(z2, loopId = "kostragr", freqVar = "ant", formula = mf2) a
mf2 <- ~region + fylke * hovedint z2 <- SmallCountData("z2") a <- PLSroundingLoop(z2, loopId = "kostragr", freqVar = "ant", formula = mf2) a
Print method for PLSrounded
## S3 method for class 'PLSrounded' print(x, digits = max(getOption("digits") - 3, 3), ...)
## S3 method for class 'PLSrounded' print(x, digits = max(getOption("digits") - 3, 3), ...)
x |
PLSrounded object |
digits |
positive integer. Minimum number of significant digits to be used for printing most numbers. |
... |
further arguments sent to the underlying |
Invisibly returns the original object.
Small count rounding via a dummy matrix and by an algorithm inspired by PLS
RoundViaDummy( data, freqVar, formula = NULL, roundBase = 3, singleRandom = FALSE, crossTable = TRUE, total = "Total", maxIterRows = 1000, maxIter = 1e+07, x = NULL, hierarchies = NULL, xReturn = FALSE, maxRound = roundBase - 1, zeroCandidates = FALSE, forceInner = FALSE, identifyNew = TRUE, step = 0, preRounded = NULL, leverageCheck = FALSE, easyCheck = TRUE, printInc = TRUE, rndSeed = 123, dimVar = NULL, plsWeights = NULL, preDifference = NULL, allSmall = FALSE, ... )
RoundViaDummy( data, freqVar, formula = NULL, roundBase = 3, singleRandom = FALSE, crossTable = TRUE, total = "Total", maxIterRows = 1000, maxIter = 1e+07, x = NULL, hierarchies = NULL, xReturn = FALSE, maxRound = roundBase - 1, zeroCandidates = FALSE, forceInner = FALSE, identifyNew = TRUE, step = 0, preRounded = NULL, leverageCheck = FALSE, easyCheck = TRUE, printInc = TRUE, rndSeed = 123, dimVar = NULL, plsWeights = NULL, preDifference = NULL, allSmall = FALSE, ... )
data |
Input data as a data frame (inner cells) |
freqVar |
Variable holding counts (name or number) |
formula |
Model formula defining publishable cells. Will be used to calculate |
roundBase |
Rounding base |
singleRandom |
Single random draw when TRUE (instead of algorithm) |
crossTable |
When TRUE, cross table in output and caculations via FormulaSums() |
total |
String used to name totals |
maxIterRows |
See details |
maxIter |
Maximum number of iterations |
x |
Dummy matrix defining publishable cells |
hierarchies |
List of hierarchies, which can be converted by |
xReturn |
Dummy matrix in output when TRUE (as input parameter |
maxRound |
Inner cells contributing to original publishable cells equal to or less than maxRound will be rounded. |
zeroCandidates |
When TRUE, inner cells in input with zero count (and multiple of roundBase when maxRound is in use) contributing to publishable cells will be included as candidates to obtain roundBase value. With vector input, the rule is specified individually for each cell. This can be specified as a vector, a variable in data or a function generating it (see details). |
forceInner |
When TRUE, all inner cells will be rounded. Use vector input to force individual cells to be rounded. This can be specified as a vector, a variable in data or a function generating it (see details). Can be combined with parameter zeroCandidates to allow zeros and roundBase multiples to be rounded up. |
identifyNew |
When |
step |
When |
preRounded |
A vector or a variable in data that contains a mixture of missing values and predetermined values of rounded inner cells. Can also be specified as a function generating it (see details). |
leverageCheck |
When TRUE, all inner cells that depends linearly on the published cells and with small frequencies
( |
easyCheck |
A light version of the above leverage checking.
Checking is performed after rounding. Extra iterations are performed when needed.
|
printInc |
Printing iteration information to console when TRUE |
rndSeed |
If non-NULL, a random generator seed to be used locally within the function without affecting the random value stream in R. |
dimVar |
The main dimensional variables and additional aggregating variables. This parameter can be useful when hierarchies and formula are unspecified. |
plsWeights |
A vector of weights for each cell to be published or a function generating it (see details). For use in the algorithm criterion. |
preDifference |
A data.frame with differences already obtained from rounding another subset of data.
There must be columns that match |
allSmall |
When TRUE, all small inner cells ( |
... |
Further parameters sent to |
Small count rounding of necessary inner cells are performed so that all small frequencies of cross-classifications to be published
(publishable cells) are rounded. This is equivalent to changing micro data since frequencies of unique combinations are changed.
Thus, additivity and consistency are guaranteed. The matrix multiplication formula is:
yPublish
=
t(x)
%*%
yInner
, where x
is the dummy matrix.
Parameters zeroCandidates
, forceInner
, preRounded
and plsWeights
can be specified as functions.
The supplied functions take the following arguments: data
, yPublish
, yInner
, crossTable
, x
, roundBase
, maxRound
, and ...
,
where the first two are numeric vectors of original counts.
When allSmall
is TRUE
, forceInner
is set to function(yInner, maxRound, ...)
yInner <= maxRound
.
Details about the step
parameter:
step
as a numeric vector is converted to three parameters by
step1 <- step[1]
step2 <- ifelse(length(step)>=2, step[2], round(step/2))
step3 <- ifelse(length(step)>=3, step[3], step[1])
After step1
steps forward, up to step2
backward steps may be performed.
At the end of the algorithm; up to step3
backward steps may be executed repeatedly.
step
when provided as a list (of numeric vectors), is adjusted to a length of 3 using rep_len(step, 3)
.
step[[1]]
is used in the main iterations.
step[[2]]
, when non-NULL
, is used in a final re-run iteration.
step[[3]]
is used in extra iterations caused by easyCheck
or leverageCheck
.
Setting step = list(0)
will result in standard behavior, with the exception that an extra re-run iteration is performed.
The most detailed setting is achieved by setting step
to a length-3 list where each element has length 3.
A list where the two first elements are two column matrices. The first matrix consists of inner cells and the second of cells to be published. In each matrix the first and the second column contains, respectively, original and rounded values. By default the cross table is the third element of the output list.
Iterations are needed since after initial rounding of identified cells, new cells are identified. If cases of a high number of identified cells the algorithm can be too memory consuming (unless singleRandom=TRUE). To avoid problems, not more than maxIterRows cells are rounded in each iteration. The iteration limit (maxIter) is by default set to be high since a low number of maxIterRows may need a high number of iterations.
See the user-friendly wrapper PLSrounding
and see Round2
for rounding by other algorithm
# See similar and related examples in PLSrounding documentation RoundViaDummy(SmallCountData("e6"), "freq") RoundViaDummy(SmallCountData("e6"), "freq", formula = ~eu * year + geo) RoundViaDummy(SmallCountData("e6"), "freq", hierarchies = list(geo = c("EU", "@Portugal", "@Spain", "Iceland"), year = c("2018", "2019"))) RoundViaDummy(SmallCountData('z2'), 'ant', ~region + hovedint + fylke*hovedint + kostragr*hovedint, 10) mf <- ~region*mnd + hovedint*mnd + fylke*hovedint*mnd + kostragr*hovedint*mnd a <- RoundViaDummy(SmallCountData('z3'), 'ant', mf, 5) b <- RoundViaDummy(SmallCountData('sosialFiktiv'), 'ant', mf, 4) print(cor(b[[2]]),digits=12) # Correlation between original and rounded # Demonstrate parameter leverageCheck # The 42nd inner cell must be rounded since it can be revealed from the published cells. mf2 <- ~region + hovedint + fylke * hovedint + kostragr * hovedint RoundViaDummy(SmallCountData("z2"), "ant", mf2, leverageCheck = FALSE)$yInner[42, ] RoundViaDummy(SmallCountData("z2"), "ant", mf2, leverageCheck = TRUE)$yInner[42, ] ## Not run: # Demonstrate parameters maxRound, zeroCandidates and forceInner # by tabulating the inner cells that have been changed. z4 <- SmallCountData("sosialFiktiv") for (forceInner in c("FALSE", "z4$ant < 10")) for (zeroCandidates in c(FALSE, TRUE)) for (maxRound in c(2, 5)) { set.seed(123) a <- RoundViaDummy(z4, "ant", formula = mf, maxRound = maxRound, zeroCandidates = zeroCandidates, forceInner = eval(parse(text = forceInner))) change <- a$yInner[, "original"] != a$yInner[, "rounded"] cat("\n\n---------------------------------------------------\n") cat(" maxRound:", maxRound, "\n") cat("zeroCandidates:", zeroCandidates, "\n") cat(" forceInner:", forceInner, "\n\n") print(table(original = a$yInner[change, "original"], rounded = a$yInner[change, "rounded"])) cat("---------------------------------------------------\n") } ## End(Not run)
# See similar and related examples in PLSrounding documentation RoundViaDummy(SmallCountData("e6"), "freq") RoundViaDummy(SmallCountData("e6"), "freq", formula = ~eu * year + geo) RoundViaDummy(SmallCountData("e6"), "freq", hierarchies = list(geo = c("EU", "@Portugal", "@Spain", "Iceland"), year = c("2018", "2019"))) RoundViaDummy(SmallCountData('z2'), 'ant', ~region + hovedint + fylke*hovedint + kostragr*hovedint, 10) mf <- ~region*mnd + hovedint*mnd + fylke*hovedint*mnd + kostragr*hovedint*mnd a <- RoundViaDummy(SmallCountData('z3'), 'ant', mf, 5) b <- RoundViaDummy(SmallCountData('sosialFiktiv'), 'ant', mf, 4) print(cor(b[[2]]),digits=12) # Correlation between original and rounded # Demonstrate parameter leverageCheck # The 42nd inner cell must be rounded since it can be revealed from the published cells. mf2 <- ~region + hovedint + fylke * hovedint + kostragr * hovedint RoundViaDummy(SmallCountData("z2"), "ant", mf2, leverageCheck = FALSE)$yInner[42, ] RoundViaDummy(SmallCountData("z2"), "ant", mf2, leverageCheck = TRUE)$yInner[42, ] ## Not run: # Demonstrate parameters maxRound, zeroCandidates and forceInner # by tabulating the inner cells that have been changed. z4 <- SmallCountData("sosialFiktiv") for (forceInner in c("FALSE", "z4$ant < 10")) for (zeroCandidates in c(FALSE, TRUE)) for (maxRound in c(2, 5)) { set.seed(123) a <- RoundViaDummy(z4, "ant", formula = mf, maxRound = maxRound, zeroCandidates = zeroCandidates, forceInner = eval(parse(text = forceInner))) change <- a$yInner[, "original"] != a$yInner[, "rounded"] cat("\n\n---------------------------------------------------\n") cat(" maxRound:", maxRound, "\n") cat("zeroCandidates:", zeroCandidates, "\n") cat(" forceInner:", forceInner, "\n\n") print(table(original = a$yInner[change, "original"], rounded = a$yInner[change, "rounded"])) cat("---------------------------------------------------\n") } ## End(Not run)
Function that returns a dataset
SmallCountData(dataset, path = NULL)
SmallCountData(dataset, path = NULL)
dataset |
Name of data set within the SmallCountRounding package |
path |
When non-NULL the data set is read from "path/dataset.RData" |
The dataset
Except for "europe6"
, "eHrc"
, "eDimList"
and "exPSD"
, the function returns the same datasets as SSBtoolsData
.
SmallCountData("z1") SmallCountData("e6") SmallCountData("eHrc") # TauArgus coded hierarchies SmallCountData("eDimList") # sdcTable coded hierarchies SmallCountData("exPSD") # Example data in presentation at Privacy in statistical databases
SmallCountData("z1") SmallCountData("e6") SmallCountData("eHrc") # TauArgus coded hierarchies SmallCountData("eDimList") # sdcTable coded hierarchies SmallCountData("exPSD") # Example data in presentation at Privacy in statistical databases