For the 1.0 release, I would concentrate on machine learning classics, essentially supervised and unsupervised learning methods (i.e. the methods for which our API is robust and well-tested). I would keep structured prediction, semi-supervised learning, active learning or anything that requires API design decisions for the 2.0 roadmap. Even if we concentrate on supervised and unsupervised learning, there are still many things to do: a neural network module, more ensemble estimators, more API consistency, the grid search API we discussed a few months ago, etc...
Modules I would consider for removal are gaussian processes, semi-supervised learning and HMMs. We could move them to a scikit-learn-bleeding-edge repository. Mathieu ------------------------------------------------------------------------------ Master HTML5, CSS3, ASP.NET, MVC, AJAX, Knockout.js, Web API and much more. Get web development skills now with LearnDevNow - 350+ hours of step-by-step video tutorials by Microsoft MVPs and experts. SALE $99.99 this month only -- learn more at: http://p.sf.net/sfu/learnmore_122812 _______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
