NEWS
mlbench 2.1-5 (2024-05-02)
- Clarify source information: the old UCI URLs no longer work.
mlbench 2.1-4
- fixed warnings in waveform.c
mlbench 2.1-2
- Moved Data-Administration.R to inst/notes
- fixed NAMESPACE problems
- added registration of C functions
mlbench 2.1-1 (2012-07-10)
mlbench 2.1-0 (2010-10-08)
- Removed dependencies on e1071 and scatterplot3d (the latter replaced by
cloud() from lattice).
- mlbench.corners() has been renamed to mlbench.hypercube().
- New function mlbench.simplex() by Manuel Eugster and Sebastian Kaiser.
- Bugfixes in the bayesclass() methods for ringnorm and threenorm
submitted by Julia Schiffer.
mlbench 2.0-0 (2010-01-21)
- Changed license of complete package to GPL-2.
- Recompressed all data sets using xz.
- Added examples to all help pages.
- Added info about UCI repository to mlbench.waveform.Rd.
mlbench 1.1-6 (2009-02-05)
mlbench 1.1-5 (2009-01-12)
- Fixed Rd bug and renamed README to LICENSE.
mlbench 1.1-4 (2008-07-02)
- Fixed documentation of S3 objects.
mlbench 1.1-3 (2007-06-01)
- Bugfix: data sets that are loaded using R code did not close the
corresponding connections.
mlbench 1.1-2 (2006-10-09)
- New data set Zoo.
- Converted the 'chas' columns of BostonHousing2 to a factor.
- Added a corrected version of PimaIndiansDiabetes as PimaIndiansDiabetes2.
- Added correct citation for UCI repository in various places.
mlbench 1.1-1 (2006-05-23)
- Add '...' argument to as.data.frame() methods for consistency
with R 2.4.0 or later.
mlbench 1.1-0 (2005-08-30)
- New artificial classification problems mlbench.shapes() and
mlbench.corners()
- Added corrected version of Boston housing data (thanks to
John Maindonald).
mlbench 1.0-1 (2005-03-07)
- Minor fixes in BostonHousing.Rd.
- Fixed a Bug in the waveform function (thanks to Karsten Lfcbke).
- Added a "ZipData: No" to the DESCRIPTION file.
- All data files are now in compressed format.
mlbench 1.0-0 (2004-02-03)
- New function mlbench.smiley().
- Version 1.0-0 is not really different from the previous release
(version 0.5-10), but we decided that it is stupid to have a
version number smaller than 1.0 for a package that is stable and
useful.