NEWS.md

In mariarizzo/kgroups: K-Groups Clustering by Energy Distance

kgroups 0.2.1

• Convert all but top level function and S3 methods to Rcpp.

kgroups 0.2.0

• Completely revised implementation.
• Much faster R version than original R version released on GitHub 2017 (last updated 2018-06-30 version 0.1.3).
• This version 0.2.0 is 100 or more times faster in benchmarks.
• Expect further improvement when compiled implementation is complete.
• Changes in arguments:
  • x can now optionally be dist object, which removed need for alpha argument.
  • alpha argument and support for it are removed - if user requires it, simply pre-compute the distances and apply the exponent to the distances in advance.
  • new optional argument clus allows starting with a given pre-initialized clustering vector.
  • method is now only "point" so the method argument and all support for moving pairs is removed from this version. To use pairs method, install version 0.1.3 from the branch.
• No dependence or reverse dependence with energy package in version 0.2.0.

kgroups 0.1.3

• Final update. This branch will not be updated after 2018-06-30. It is replaced by a completely rewritten R version. Some features will no longer be supported; other features will be added. The new code is faster.
• Fixed a bug in OnePass.R at updating step for within cluster (bug inherited from original code of S. Li).
• other minor internal changes

kgroups 0.1.2

• Fixed iteration counter on 2018-05-19
• Added the NEWS file

kgroups 0.1.1

• Initial release on GitHub 2017-11-14 repackaging and documenting by M. Rizzo of code developed by Songzi Li in 2014.

mariarizzo/kgroups documentation built on Aug. 29, 2019, 11:54 a.m.