-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The NaN-toolbox has been extended by several important features and functions. Most notable are:
- - classification and machine learning are supported - - support of data weighting - - use of extended accuracy in sumskipnan_mex and covm_mex in order to reduce possible round-off errors. The changes justify a change of the major version number, which becomes now NaN-toolbox v2.0 Specifically, the functions *train_sc()* and *test_sc.m* provide a number of different classification and machine learning methods, including Linear (LDA), Quadratic (QDA/MDA) and regularized discriminant analysis (RDA), Naive Bayes classifier (NBC), augmented NBC (aNBC), regression and partial least squares (REG/PLS), WienerHopf, some sparse classifiers, as well as an interface to several SVM classifiers are implemented. *classify.m* resembles some functionality of the corresponding functions from the ML stats-toolbox. However, it is recommended to use *xval.m* instead, which is a wrapper for using cross-validation. *kappa.m* estimates the Cohen's kappa coefficient as well as several other evaluation criteria based on the concordance matrix between the original and the classified data labels. Alois -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkpMtO8ACgkQzSlbmAlvEIi/SwCeIIFkkBMU5scmdet2PpnJt453 wL0An2b/wq8Rh3ZC9j1F78aXiGjB9XQI =1KUF -----END PGP SIGNATURE----- ------------------------------------------------------------------------------ _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
