Deductive Correction, Deductive Imputation, and Deterministic Correction

A collection of methods for automated data cleaning where all actions are logged.


version 1.3.7

  • Small updates to conform to CRAN policies.

version 1.3-5

  • Small updates to conform to CRAN policies.

version 1.3-4

  • bugfix in logging of correctWithRules (factors were incorrectly logged).

version 1.3-1

  • abs is now an allowed symbol in deterministic corrections (see ?correctionRules)
  • small documentation updates: the DL-distance used in correctTypos is the restricted DL-distance.
  • bugfix in correctSigns: not all swap-options were found when edits involving <= (or >=) are involved (thanks to Sander Scholtus).
  • bugfix deterministic imputation: wrong logging when multiple rules were applied to a record (thanks to Jeroen Pannekoek).

version 1.3-0

  • Added deterministic imputation (correction) feature, see ?imputationRules

version 1.2-1

  • Added 'round' option for deduImpute.editmatrix (defaults to true)
  • bugfix in correctSigns: some correction opportunities were missed because of ignored constant vector in constraints.

version 1.2-0

  • Discussion papers used to be included as vignette. Now only as pdf in inst/doc
  • deducorrect object gains 'call' attribute.
  • deduImpute now sets data to NA according to adapt (should be more intuitive)
  • deduImpute accepts 'editset'. Imputations causing new violations are reverted afterwards
  • correctRounding, correctTypos, correctSigns now S3 generic and accept 'editset'. Corrections causing new violations are reverted afterwards.
  • deduImpute now accepts data.frames with more variables than specified in editset (-matrix, -array)
  • bugfix: counting nr. of imputations in deduImpute.editmatrix in presence of 'adapt' argument
  • BREAKING CHANGE: Removed deprecated argument 'Q' from 'correctRounding'
  • BREAKING CHANGE: Argument 'R' of correctRounding is renamed 'E' for consistency reasons

version 1.1-3

  • Solved problem in vignette, caused by updates in editrules.

version 1.1-2

  • Discussion paper included in 'doc' directory of package
  • Solved bug in deducorrect object: rownames of $corrections were sometimes duplicated

version 1.1-1

  • Updates on vignette, index.html of online manual and documentation.
  • Solved possible clash with plyr: both editrules and plyr publish a reduce function.

version 1.1-0

  • Deductive imputation methods for categorical data: deduImpute, deductiveLevels
  • Deductive imputation methods for numerical records: deduImpute, solSpace/imputess, deductiveZeros
  • Solved corner case: constructing deducorrect object when nothing was corrected failed.

version 1.0-2

  • correctSigns has more thorough argument checking. Variables in flip or swap, not occuring in rules are removed.
  • Test for unimodularity of editmatrix in correctRounding can optionally be turned off.
  • fixed a bug in correctRounding causing inequality violations in certain cases (Thanks to Kenneth Chin-A-Fat)
  • passing argument Q to correctRoundings now emits error

version 1.0-1

  • fixed bug in correctSigns causing inequality violations in certain cases.

version 1.0-0

  • discussion paper published and added to package documentation

version 0.9-6

  • Shrunk included papers with tools::compactPDF()
  • Small changes to fix warnings and errors after update of editrules and R2.13-patched
  • Vignette finished

version 0.9-5

  • vignette nearly finished
  • Package now depends on editrules version >= 0.6
  • Changed getUsername so it uses

version 0.9-4

  • Fixed a bug in correctTypos: all solutions were flagged as partial.
  • Fixed a bug in damerauLevenschtein distance

version 0.9-3

  • vignette virtually complete
  • made detection of username more robust
  • correctRounding does not allow generation of new inequality violations, compatible with correctTypos.
  • correctTypos now checks solution suggestions for generating new inequality violations.
  • set variable Q in correctRounding to deprecated
  • fixed bug in isTotallyUnimodular
  • old.correctSigns now removed
  • added index.html for vignettes

version 0.9-2

  • some work on vignette
  • deducorrect object how has "user" slot
  • correctSigns has new, easier interface and can take inequalities into account.
  • correctSigns now breaks of a 1e5 sign flip combinations - can be adjusted by setting "maxCombinations"
  • solved bug causing correctSigns to crash if no records at all were fixed.
  • correctRounding failed when running on R 2.10, is fixed.

version 0.9-1

  • Fixed a bug in correctTypos: when a record was invalid but no sensible corrections could be made, the function returned an error
  • Smashed a bug in correctSigns: error extracting constant for editmatrix containing equalities and inequalities

version 0.9

  • initial version

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


1.3.7 by Mark van der Loo, 4 years ago

Browse source code at

Authors: Mark van der Loo , Edwin de Jonge , and Sander Scholtus

Documentation:   PDF Manual  

Task views: Official Statistics & Survey Methodology

GPL-3 license

Imports stats, utils, methods

Depends on editrules

See at CRAN