Release Notes: Logistic regression was added. Demos are included in the source and documentation. Cluster agreement metrics were added. An nfoldcrossvalidation bug when using the origins parameter was fixed.
Release Notes: New features: unsupervised (1-class) kernel density modeling, a weights option to some learners, stump learner, and Adaboost. A fix for when SDA returns empty.
Release Notes: A fix was included for 64-bit machines. Functions in measures.py have a new interface.
Release Notes: Random forest learners were added. Decision trees were sped up by 20 times. Gridsearch is much faster since it finds an optimum without computing all folds.
Release Notes: A missing file that prevented installation was included.
Release Notes: The dependency on scipy was minimized. Self-organizing maps were added. An important bug was fixed in repeated_kmeans. Feature selection is faster.
Release Notes: This release fixes sparse non-negative matrix factorization. It has a mean grouped classifier, updates the multi-classifier to a newer interface, and has better testing and documentation.
Release Notes: The dependency on scipy.weave was removed. The namespace was made flatter. K-means was made faster. Affinity propagation was borrowed from scikits-learn and slightly improved. pdist() was added.
Release Notes: Bugfixes, faster kmeans, and improvements to nfoldcrossvalidation.