It's been a long time, but there is a new release. 0.1.6 has been release quietly a while ago, so this will be number 0.1.7.
Things that are new or changed since the last announcement:
- Spectral clustering algorithms are available in SVN, will be included in the main release further down the line.
- A whole lot more unit tests
- Bayes classifiers
- More tutorials
- Improved sampling strategies
- fixed getWeights for sparse data sets
- improved Dynamic Time Warping implementation
- upgraded most dependencies to more recent versions (libsvm 2.9.1, ajt 2.9)
- fixed bug in Kmeans that entered an infinite loop in a particular case
- fixed bug in normalization for attributes that have no range
- Fixed a whole bunch of bugs contributed through the Sourceforge ticketing system
- Removed OPTICS implementation until further notice because it is incomplete