On 06/04/2014 04:21 PM, Olivier Grisel wrote: > Hi all, > > I am thinking about cutting a 0.15 branch as soon as the PR on > multilabel input format is merged: > > https://github.com/scikit-learn/scikit-learn/pull/2657 > > and release a 0.15.0b1 right after to test packaging under various > platforms (it should be good as I recently got good windows builds > under 2.7 and 3.4 and the MSVC compiler). > > Then work on fixing bugs to release 0.15.0b2 next week and 0.15.0 > final in 2 weeks. > > Merging new features can still happen in master during that period. Only > bugfixes are to be cherrypicked to the 0.15 branch. > > I would really like to find the time to finish investigating how to > best fix the minibatch k-means cluster reassignment during the beta / > bug fixing period but I don't want to delay the release process any > further as I am already 7 months late on my initial schedule :P > Actually I don't care so much about the strategy as about the fixes being merged. The current implementation is pretty buggy.
Also, I'd really like the fix for train_val_split to be merged. Having cross-validation work on text data, but train_val_split not (memory explosion) seems pretty bad to me. Sorry I haven't been working on these :-/ ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/NeoTech _______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
