Software and datasets to support 'Modern Applied Statistics with S', fourth edition, by W. N. Venables and B. D. Ripley. Springer, 2002, ISBN 0-387-95457-0. This file documents software changes since the third edition. - no copying of datasets even in R. - model.frame method for multinom (even in R). - nnet now uses the C interface to optim. - nnet.Hess has been renamed nnetHess. - vcov.multinom now computes the Hessian analytically (thanks to David Firth). - predict methods for multinom, nnet now check newdata types - model.frame.multinom now looks for the environment of the original formula - multinom has a new `model' argument defaulting to TRUE. - the multinom methods for add1, dropterm and anova now check for changes in the number of cases in use caused e.g. by na.action=na.omit. - added confint() method for multinom. - added logLik() method for multinom. - summary() for multinom now defaults to correlation=FALSE. - nnet() reports on 'convergence'. - confint.multinom() works better with a non-default 'parm'. - multinom() and nnet(softmax=TRUE) give an explicit error message for one-category responses. - the loglik() method for multinom() returns an "nobs" attribute. - vcov() on a "multinom" object works for fits with na.action = "na.exclude".