EOV

EOV (Ensemble Of Voters) is a data mining application that is optimized for classification tasks. It implements an ensemble of voting decision stumps and a naïve Bayesian network. EOV is currently undergoing a major rewrite that will make it more flexible and powerful, give it a more polished command-line interface, and will ultimately incorporate full Bayesian network learning into its repertoire of algorithms.

Some related applications are: C5.0, a data mining application that uses decision trees; and SVMlight, a data mining application that uses support vector machines. EOV attempts to maintain a certain level of compatibility with these two systems and will, in coming releases, be working to increase that level.

The development of EOV is supported in part by a training grant from the National Library of Medicine (Grant No. 5T15LM007359).

EOV is distributed under the GNU General Public License.

Version 1.0fb-alpha1 (Source) - This is a version of EOV that has initial support for full Bayesian-structure learning. Note, that this version only supports Bayesian inference and not an ensemble of decision stumps.

Version 1.01 (Patch) - A patch for Version 1.0 that adds support to output SVMlight .train and .test files after feature selection has occurred.

Version 1.0 (Source) - This is the initial public release. This is considered "research-grade" code - most options are specified as compiler flags in the file datapoints.h, not command-line options.