SmallCountRounding 1.1.0

• Added a check to ensure that at least one of dimVar, hierarchies, or formula is specified.
  • This is a breaking change that may affect previous code.
  • Previously, if all were unspecified, dimVar was automatically generated from the remaining columns.
  • While this behavior was correctly implemented, it often stemmed from user input errors and could lead to unexpected behavior or crashes.
  • This change now requires explicit input, making the function more robust and reducing the risk of user errors.
• Improved support for tibble and data.table input (parameter data).
  • Input is now explicitly coerced to a data frame using as.data.frame() where necessary to ensure consistent behavior.
  • When preAggregate is TRUE and aggregatePackage is "data.table", the use of as.data.frame() is skipped to avoid unnecessary back-and-forth conversion of data.table objects, preserving efficiency.
  • Applies to PLSrounding() and its wrappers.
• Note new hierarchy possibilities due to the new version of the SSBtools package (version 1.6.0).

  • Output from functions like get_klass() in the klassR package or hier_create() in the sdcHierarchies package can now be used directly as input. Example of usage:

     a <- get_klass(classification = "24")
     b <- hier_create(root = "Total", nodes = LETTERS[1:5])
     mydata <- data.frame(tree = sample(a$code[nchar(a$code) > 1], 200, replace = TRUE), 
                          letter = LETTERS[1:5])
     PLSroundingPublish(mydata, roundBase = 5, hierarchies = list(tree = a, letter = b)) 

  • New possibilities for working with both formulas and hierarchies are now available through the map_hierarchies_to_data() function.

  • Improved functionality for combining formulas with the Formula2ModelMatrix() parameter avoidHierarchical = TRUE, thanks to the new total_collapse() function which can be applied to output.

SmallCountRounding 1.0.8

• FormulaSelection() now works with the output from PLSrounding().
  • The output dataset corresponding to a restricted part of the input formula can now be easily retrieved.
  • See the examples in the documentation.
• extend0 is new parameter to PLSrounding(), enabling data to be automatically extended by zero frequency rows.
  • This is relevant when zeroCandidates = TRUE.
  • The old parameter in PLSroundingFits() has been renamed from extend0 to extend0Fits. Code that used the old parameter will now behave differently.
  • Note that extend0 and extend0Fits can now be specified in more advanced ways beyond just TRUE/FALSE.
• Improvements to the step parameter, which can be passed to PLSrounding() and is documented in the underlying function RoundViaDummy():
  • A bug that could cause a hang when using step has been fixed.
  • The step parameter can now be specified as a vector for greater control.
  • Additionally, it can be provided as a list to trigger a final re-run iteration.
  • The step parameter can significantly impact performance on large datasets. For example, using step = list(100) may be a useful approach.
• Due to updates in the SSBtools package (version 1.5.4), it is now meaningful to include NA’s in the grouping variables.
  • Note the parameter NAomit to SSBtools::Formula2ModelMatrix():
    • When TRUE, NAs in the grouping variables are omitted in output and not included as a separate category.
    • This parameter can be input to PLSrounding() and its wrappers.
  • aggregateNA is new parameter to PLSrounding():
    • Whether to include NAs in the grouping variables while preAggregate.
    • Needs to be TRUE (default) to utilize the above NAomit parameter.
• Due to updates in the SSBtools package (version 1.5.4), where data.table is now listed under Suggests, some functionality can be speeded up.
  • Set the new parameter aggregatePackage to "data.table" to utilize this possibility.
    • aggregatePackage is parameter to PLSrounding() and its wrappers.
    • Also note the related new parameters aggregateBaseOrder.

SmallCountRounding 1.0.5

• Minor updates with no changes in functionality
  • Changed package license to MIT, in accordance with the policy at Statistics Norway.
  • Some technical changes in documentation to comply with standards.

SmallCountRounding 1.0.3

• Workaround for old R versions where the isFALSE function is not defined.

SmallCountRounding 1.0.2

• Improved behavior of the identifyNew parameter when the maxRound parameter is used.
  • New description of the identifyNew parameter: When TRUE, new cells may be identified after initial rounding to ensure all rounded publishable cells equal to or less than maxRound to be roundBase multiples. Use NA for the a less conservative behavior (old behavior). Then it is ensured that no nonzero rounded publishable cells are smaller than roundBase. When maxRound is default, there is no difference between TRUE and NA.

SmallCountRounding 1.0.0

• New function, PLSroundingLoop: PLSrounding on portions of data at a time.
  • The runs are coordinated by using preliminary differences as input for the next run (parameter preDifference)
• Parameters zeroCandidates, forceInner, preRounded and plsWeights can now be specified as functions.
  • These cannot be supplied as vectors in PLSroundingLoop.
• New parameter, allSmall.
  • When TRUE, all small inner cells (<= maxRound) are rounded. A simplified alternative to specifying forceInner.
• Adaption needed after Matrix ver. 1.4-2 (not a user-visible change)

SmallCountRounding 0.9.0

• New function, PLSroundingFits, for post-processing to expected frequencies
  • Expected inner cell frequencies are generated by iterative proportional fitting
• plsWeights is new parameter to RoundViaDummy (and PLSrounding)
  • A vector of weights for each cell to be published or a function generating it. For use in the algorithm criterion.

SmallCountRounding 0.8.0

• Now, microdata input is allowed. This is due to
  • Allowing empty freqVar in input.
  • The new parameter preAggregate: When TRUE, the data will be aggregated beforehand within the function by the dimensional variables.
• It is possible to avoid handling hierarchical variables when using the formula interface.
  • This is a consequence of parameter avoidHierarchical to Formula2ModelMatrix in the SSBtools package.

SmallCountRounding 0.7.0

• Now, a random generator seed is used locally within the function without affecting the random value stream in R.
  • Handled by rndSeed, a new parameter to RoundViaDummy (and PLSrounding).
  • By default, rndSeed = 123. This means that repeated runs with equal input will result in equal output.
  • To get back the old behaviour of the function, set rndSeed to NULL.
• Possible to return a single data frame: "inner" or "publish".
  • Handled by output, a new parameter to PLSrounding.
  • New wrapper functions, PLSroundingInner and PLSroundingPublish.
• dimVar is new parameter to RoundViaDummy and PLSrounding
  • The main dimensional variables and additional aggregating variables. This parameter can be useful when hierarchies and formula are unspecified.

SmallCountRounding 0.6.0

• preRounded is new parameter to RoundViaDummy (and PLSrounding)
  • A mixture of missing values and predetermined values of rounded inner cells

SmallCountRounding 0.5.0

• Formula combined with hierarchies is now possible
  • This is a consequence of the function HierarchiesAndFormula2ModelMatrix in the SSBtools package
• leverageCheck and easyCheck are new parameters to RoundViaDummy
  • This provides protection against possible disclosure of small numbers by linear relationships (difference attack)
  • The function Reduce0exact in the SSBtools package is utilised
• printInc is new parameter to PLSrounding and RoundViaDummy
  • Whether to print iteration information to the console
• Possible to specify removeEmpty=TRUE to omit empty combinations
  • Parameter to Hierarchies2ModelMatrix and HierarchiesAndFormula2ModelMatrix in the SSBtools package
• The parameter inputInOutput is also mentioned in the RoundViaDummy documentation
  • Parameter to same functions as above
  • Can be used to specify whether to include codes from input
• A vignette, “Introduction to ‘SmallCountRounding’”, is included

SmallCountRounding 0.4.0

• Last version before any news