Changes in version 2025-10-21                      

  - In the user-interface, the "significance" of MMDs are now decided
    using Sjovold's rule-of-thumb (the MMD value must be greater than
    twice its standard deviation). The p-values are not used for that
    purpose anymore, since their validity is still highly discussed in
    the literature, and relies on often unrealistic hypotheses.
  - In the same spirit, p-values have now been totally removed from the
    package. {AnthropMMD} now only provides ratios MMD/sd(MMD), as in
    Bertsatos' Octave script (see Bertsatos & Chovalopoulou, 2016).

Other minor changes

  - More unit tests have been added, and some unit tests have been
    rewritten.
  - A codemeta.json file has been added to provide a better structure
    for the package's metadata.

                     Changes in version 2023-11-29                      

Other changes

  - Fixed man page of package (version mismatch)
  - New reference added in man pages

                     Changes in version 2023-07-28                      

Other changes

  - Fixed encoding of LICENSE file.

                     Changes in version 2023-06-15                      

Bug fix

  - The internal function mmd_resample() now handles edge cases where
    resampling problems could arise.
  - Fixed formula for the computation of sample sizes in mmd_resample().

                     Changes in version 2023-03-06                      

Requirements

  - AnthropMMD now depends on R >= 4.1.0.
  - New dependencies on third-party R packages ({dplyr}, {MASS},
    {rlang}) have been added.

New features

  - The function mmd() now has another argument, correct (which defaults
    to TRUE), which allows the user to specify whether the correction
    for small sample sizes should be applied.
  - A bootstrap method described by Fidalgo et al. (2022) is now
    available through the function mmd_boot(). It can be used from the
    command line interface only; its implemention in the graphical user
    interface is not planned (at least in a near future).

Other changes

  - The code of various functions (in particular binary_to_table()) has
    been refactored, and should now run faster.
  - Updated package vignette.

                     Changes in version 2020-07-17                      

Incompatible changes in the API

  - A new generic function, plot(), is added. It replaces plot_mmd().
    The old plot_mmd() is not exported anymore.

New features

  - p-values are now implemented in mmd().

Other changes

  - Improved code readability.

                     Changes in version 2019-07-16                      

Minor changes

  - Fixed warning when the angular transformation is not specified in
    mmd() function.
  - Fixed typos in package vignette and documentation files.

                     Changes in version 2019-07-05                      

Change in dependencies

  - Due to some new features implemented in AnthropMMD, the package now
    depends on R >= 3.5.0.
  - AnthropMMD now imports {plotrix}.
  - AnthropMMD now suggests {covr}, {knitr}, {rmarkdown} and {testthat}.

Change of License

  - AnthropMMD is now distributed under CeCILL 2.1 license (instead of
    GPL 3).

User visible changes

  - AnthropMMD can now be used by command lines, and not only as an
    R-shiny application. The main goal is to make AnthropMMD suitable
    for reproducible research.
  - An example data file, toyMMD, is now available.
  - A vignette is now available.
  - Improved display of group labels on the MDS plots.
  - Some small improvements in documentation files.

Other changes

  - A large majority of comments in the R source files have been
    translated from French into English.
  - The reliability of future updates has been improved by implementing
    unit tests; and a CI pipeline has been set on the GitLab repo.
  - The indentation style in .R and .Rd files now follows the R
    standards.

Bug fixes

  - The Shiny app does not crash anymore if the user specifies a wrong
    field separator.
  - Fix for the computation of Spearman's measure of agreement for MDS,
    where distances were erroneously counted twice. Although the impact
    was generally negligible, this resulted in a small overestimation of
    the quality of agreement.

                     Changes in version 2019-03-18                      

  - First release on GitLab
  - Added LICENSE, NEWS and README files
  - Change of GPL license (GPL3 instead of GPL2)
  - Updated DESCRIPTION file