Changes in Verions 1.5-18 o Improve DESCRIPTION install metadata. Changes in Version 1.5-17 o tune() now also returns a dispersion measure of all training samples. Bootstrap is done *with* replacement. o tune.svm now also accepts the epsilon-parameter Changes in Version 1.5-16 o write.svm now also stores the scaling information for the dependent variable o data sets Glass, HouseVotes84, and Ozone removed (are in package mlbench) o merged help pages for naiveBayes and predict.naiveBayes Changes in Version 1.5-15 o Bug in NAMESPACE-file fixed (conditional import from utils failed in R 2.3.1). Changes in Version 1.5-14 o predict.naiveBayes sped up o Bug fix in plot.svm (error in case of training categories without predictions) o methods now added to `Suggests:', and grDevices to imports Changes in Version 1.5-13 o Bug fix: sparse handling was broken since 1.5-9 Changes in Version 1.5-12 o update to libsvm 2.81 o laplace smoothing added to naiveBayes() Changes in Version 1.5-11 o tune: allow list of vectors as tune parameter range so that class.weights in svm-models can be tuned. o better default color palette for plot.tune(). o New function probplot() for probability plots. Changes in Version 1.5-10 o Bug fix: class probability prediction was broken since 1.5-9 Changes in Version 1.5-9 o tune() now returns the split indices into training/validation set. Information added about cross validation. o plot.svm(): wrong labeling order in levels fixed. o predict.svm now adds row numbers to predictions, and correctly handles the na.action argument using napredict() Changes in Version 1.5-8 o Update to libsvm 2.8 (uses a faster optimization algorithm) Changes in Version 1.5-7 o read.matrix.csr() did not work correctly with matrix-only objects. o svm(): Fixed wrong labeling for predicted decision values and probabilities in case of a Class factor created from a non-ordered character vector. Changes in Version 1.5-6 o cmeans() is substantially enhanced, with a complete rewrite of the underlying C code. It is now possible to specify case weights and the relative convergence tolerance. For Manhattan distances, centers are correctly computed as suitably weighted medians (rather than means) of the observations. The print method for fclust objects is now more in parallel with related methods, and registered in the name space. Changes in Version 1.5-5 o read.octave() is now deprecated in favor of a substantially enhanced version in package foreign for reading in files in Octave text data format. Changes in Version 1.5-4 o Use lazy loading. Changes in Version 1.5-3 o New arguments in plot.svm for customizing plot symbols and colors o Fix of broken code in plot.svm for the `fill=FALSE' (non-default) case Changes in Version 1.5-2 o Fixed memory leak in svm Changes in Version 1.5-1 o Fixed C++ style comments Changes in Version 1.5-0 o Example for weighting added in svm() help page o upgrade to libsvm 2.6: support for probabilities added Changes in Version 1.4-1 o NaiveBayes() is more accurate for small probabilities o call is more sensible in tune(), tune.foo(), and best.foo() objects o `control' parameter of tune() changed to `tunecontrol' to solve name space conflict with training methods using `control' themselves o new function matchControls() o fixed a bug in bclust() triggered when a cluster had only one center Changes in Version 1.4-0 o adjusted to restructering of R base packages o added a NAMESPACE file o Function `write.svm' now also creates a file with scaling information Changes in Version 1.3.16: o Small bug fixes in predict.svm() and plot.svm() o Function `write.svm' added which saves models created with svm() in the format libsvm can read. Changes in Version 1.3.15: o Bug fix in plot.svm(): non-SVs had wrong colors o data sets `Ozone' and `Glass' added Changes in Version 1.3.14: o Several Docu bug fixes (plot.bclust, impute, stft, svm.formula, svm.default) o upgrade to libsvm 2.5. New feature: predict.svm() optionally returns decision values for multi-class classification o svm-vignette gave warnings due to rank deficiency in Ozone data o naiveBayes() now also supports metric predictors, and the standard interface. Changes in Version 1.3.13: o Bug fixes in svm: - Prediction of 1 single observation gave an error - Only k instead of k*(k-1)/2 rho coefficients have been returned by svm (k number of classes), having caused nonsensical results for k > 3. o The `svmdoc' file in inst/doc now is a vignette. Changes in Version 1.3-12: o The x argument of cmeans() and bclust() is now automatically coerced to a matrix. o Started tests directory o New method: `naiveBayes' classifier for categorical predictors o optimization of `read.matrix.csr' which used to be rather slow o Bug fixes for the `svm' interface: when the data included categorical predictors, the scaling procedure did not only affect the metric variables, but also the binary variables in the model matrix. o Function `scaclust removed'. Bug has to be fixed. Changes in Version 1.3-10: o Now supports libsvm 2.4 Changes in Version 1.3-9: o rdiscrete() is now simply a wrapper for sample() and provided for backwards compatibility only. o Minor bug fixes in `svm' and `tune' (mostly interface issues). New plot function for objects of class `svm' working for the 2d-classification case. Changes in Version 1.3-7: o svm() now supports the `matrix.csr' format, as handled by the `SparseM' package. Predictors and response variable (if numeric) are scaled per default. o A new `plot' function for `svm' objects visualizes classification models by plotting data and support vectors in the data input space, along with the class borders. o A new generic `tune' function allows parameter tuning of arbitrary functions using, e.g., boot strapping, or cross validation. Several convenience wrappers (e.g., for svm, nnet, and rpart) do exist. Changes in Version 1.3-3: o Bug fixes in various bclust routines: stop() if required packages are not found o svm() now interfaces LIBSVM 2.35 which is a bug fix release. A call with invalid parameters now no longer causes R to be terminated, and the C(++) code became completely silent. o Bugs fixed in fclustIndex function and print.fclust. Changes in Version 1.3-1: o Functions rmvnorm and dmvnorm for multivariate normal distributions have been moved to package mvtnorm. o Bug fixes in print.fclust and fclustIndex. o fixed floyd.c (ANSI C pedantic warnings) Changes in Version 1.2-1: o Bug fixes in cmeans.c, cshell.c and scaclust.c (R header files included and unused variables removed) o Bug fixes in Rsvm.c and svm.R (incomplete list of returned Support Vectors). o Encapsulate kmeans call in bclust() in a try construct, because kmeans gives an error when a cluster becomes empty (which can happen for almost every data set from time to time). Changes in Version 1.2-0: o Added functions for bagged clustering, see help(bclust). o read.pnm() and write.pgm() have been removed from e1071, much improved versions can now be found in the new package pixmap. o Lots of documentation updates and bugfixes. o Support Vector Machine interface now upgraded to libsvm V. 2.31 featuring: - Multi-Class Classification - weighting of classes for C-classification (for asymmetric sample sizes) - nu-regression - Formula Interface - k-fold cross-validation In addition, an introductory article is provided in directory `docs/' (svmdoc.pdf). o classAgreement now features an option to match factor levels o updated API design for the fuzzy clustering functions (cmeans, cshell, scaclust). Documentation updates and function name changes (cmeanscl -> cmeans, validity.measures -> fclustIndex)