All the data sets are available, except faithful and tree which are
already in R.

Additional datasets to work our examples are:

VA, heart

The following are fully operational:

ginv		generalized inverse
write.matrix	write matrix or numeric data frame
eqscplot	equally-scaled plots

mvrnorm 	simulate from a multivariate normal distribution
huber		Huber location with MAD scale
hubers		Huber proposal 2
truehist	histogram plots
nclass.scott	choose number of histogram classes
nclass.FD	choose number of histogram classes
hist.scott	histogram with other bin choices
hist.FD 	histogram with other bin choices
frequency.polygon
nclass.freq	choose number of classes for frequency polygon
ucv, bcv	cross-validation functions for density
width.SJ	bandwidth choice for density.
kde2d		2D kernel density estimation

boxcox		estimate Box-Cox transformation
logtrans	estimate shift in log-transformations
stdres		standardized residuals for linear models
studres		Studentized residuals for linear models
vcov		variance-covariance summary for linear, glm and 
		   non-linear regression fits
contr.sdif	contrasts function for successive differences
fraction	methods for operating on rational fractions

rlm		robust linear model fitting (by Huber M-estimator)
lqs		resistant regression
addterm		try all one-term additions to a model
dropterm	try all one-term deletions from a model
stepAIC		stepwise AIC minimization
lm.gls		generalized least squares fit
lm.ridge	ridge regression

dose.p		function for LD50-like fits
neg.bin 	negative binomial family with fixed theta
anova.negbin	functions for full negative binomial family
glm.convert
glm.nb
negative.binomial
rnegbin
theta.md
theta.mm
loglm		Object-oriented wrapper for loglin() (from complements)
polr		proportional odds logistic regression
gamma.shape	MLE of shape parameter for GLM fit with gamma family
gamma.dispersion		(from complements)


deviance.nls	deviance function for nls fits
plot.profile	plot method for profiles
pairs.profile	pairs method for profiles
profile.glm	profiles for glm objects

lda		linear discrimination
predict.lda	predict function for lda
qda		quadratic discrimination
predict.qda	predict function for qda
sammon		Sammon non-linear mapping
isoMDS		isotonic multidimensional scaling
Shepard 	Shepard plot for isoMDS
cov.trob 	robust estimation of multivariate location and scatter
corresp 	correspondence analysis
mca		multiple correspondence analysis (complements)


moved to base R:
max.col		find largest column for each row of a matrix,
		breaking ties at random
digamma, trigamma	first and second derivatives of log Gamma
		(even of complex argument).

IQR is in base R
lqs, cov.rob are in package lqs in R
biplot.princomp is in package mva in R
ppr is in package modreg in R
cpgram is in package ts in R

The following depend on facilities not in R and so do not work:

Choleski	Choleski decompositions for Matrices
histplot	Trellis histogram plots
rms.curv	curvature measures for non-linear regression
deriv3		symbolic differentiation
D, make.call	extended functions
vcov.nlregb	vcov methods for nlregb objects
vcov.nlminb	vcov methods for nlminb objects


BDR <ripley@stats.ox.ac.uk> 2000/08/09

