Robust (or "resistant") methods for statistics modelling have been
available in S from the very beginning in the 1980s; and then in R in
package
stats.
Examples are
median(),
mean(*, trim =. ),
mad(),
IQR(),
or also
fivenum(), the statistic
behind
boxplot()
in package
graphics)
or
lowess()
(and
loess()) for robust
nonparametric regression, which had been complemented
by
runmed()
in 2003.
Much further important functionality has been made available in
recommended (and hence present in all R versions) package
MASS
(by Bill Venables and Brian Ripley, see
the
book
Modern Applied
Statistics with S
).
Most importantly, they provide
rlm()
for robust regression and
cov.rob()
for
robust multivariate scatter and covariance.
This task view is about R addon packages providing newer or faster,
more efficient algorithms and notably for (robustification of) new models.
Please send suggestions for additions and extensions to the
task view maintainer
.
An international group of scientists working in the field of robust
statistics has made efforts (since October 2005) to coordinate several of
the scattered developments and make the important ones available
through a set of R packages complementing each other.
These should build on a basic package with "Essentials",
coined
robustbase
with (potentially many) other packages
building on top and extending the essential functionality to particular
models or applications.
Further, there is the quite comprehensive package
robust, a version of the robust library of SPLUS,
as an R package now GPLicensed thanks to Insightful and Kjell Konis.
Originally, there has been much overlap between 'robustbase'
and 'robust', now
robust
depends
on
robustbase, the former providing convenient routines for
the casual user where the latter will contain the underlying
functionality, and provide the more advanced statistician with a
large range of options for robust modeling.
We structure the packages roughly into the following topics, and
typically will first mention functionality in packages
robustbase
and
robust.

Regression (Linear, Generalized Linear, Nonlinear Models,
incl. Mixed Effects)
:
lmrob()
(robustbase) and
lmRob()
(robust) where the former uses the latest of the
fastS algorithms and heteroscedasticity and autocorrelation corrected
(HAC) standard errors, the latter makes use of the MS algorithm of
Maronna and Yohai (2000), automatically when there are factors
among the predictors (where Sestimators (and hence MMestimators)
based on resampling typically badly fail).
The
ltsReg()
and
lmrob.S()
functions
are available in
robustbase, but rather for comparison
purposes.
rlm()
from
MASS
had been the first widely
available implementation for robust linear models, and also one of
the very first MMestimation implementations.
robustreg
provides very simple Mestimates for linear
regression (in pure R).
Note that Koenker's quantile regression package
quantreg
contains L1 (aka LAD, least absolute deviations)regression as a
special case, doing so also for nonparametric regression via
splines.
Quantile regression (and hence L1 or LAD) for mixed effect models,
is available in package
lqmm, whereas an
MMlike
approach for robust linear
mixed effects
modeling
is available from package
robustlmm.
Package
mblm
's function
mblm()
fits
medianbased (TheilSen or Siegel's repeated) simple linear models.
Package
TEEReg
provides trimmed elemental estimators for
linear models.
Generalized linear models (GLMs) are provided both via
glmrob()
(robustbase) and
glmRob()
(robust),
where package
robustloggamma
focuses on generalized log
gamma models.
Robust ordinal regression is provided by
rorutadis
(UTADIS).
Robust Nonlinear model fitting is available through
robustbase
's
nlrob().
multinomRob
fits overdispersed multinomial regression
models for count data.
rgam
and
robustgam
both fit robust GAMs,
i.e., robust Generalized Additive Models.
drgee
fits "Doubly Robust" Generalized Estimating Equations (GEEs)
complmrob
does robust linear regression with compositional data as covariates.

Multivariate Analysis
:
Here, the
rrcov
package which builds ("
Depends
")
on
robustbase
provides nice S4 class based methods,
more methods for robust multivariate variancecovariance estimation,
and adds robust PCA methodology.
It is extended by
rrcovNA, providing robust multivariate
methods for
for incomplete
or missing (
NA) data, and by
rrcovHD, providing robust multivariate methods for
High Dimensional
data. High dimensional data with an
emphasis on functional data are treated robustly also by
roahd.
Specialized robust PCA packages are
pcaPP
(via
Projection Pursuit),
rpca
(incl "sparse")
and
rospca.
Historically, note that robust PCA can be performed by using standard
R's
princomp(), e.g.,
X < stackloss; pc.rob < princomp(X, covmat= MASS::cov.rob(X))
Here,
robustbase
contains a slightly more flexible
version,
covMcd()
than
robust
's
fastmcd(), and similarly for
covOGK().
OTOH,
robust
's
covRob()
has automatically chosen
methods, notably
pairwiseQC()
for large dimensionality p.
Package
robustX
for experimental, or other not yet
established procedures, contains
BACON()
and
covNCC(), the latter providing the
neighbor variance estimation (NNVE) method of Wang and Raftery (2002),
also available (slightly less optimized) in
covRobust.
RobRSVD
provides a robust Regularized Singular Value Decomposition.
mvoutlier
(building on
robustbase) provides
several methods for outlier identification in high dimensions.
GSE
estimates multivariate location and scatter in the presence of missing data.
FRB
performs robust inference based on
F
ast
and
R
obust
B
ootstrap on robust estimators, including
multivariate regression, PCA and Hotelling tests.
RSKC
provides
R
obust
S
parse
K
means
C
lustering.
robustDA
for
robust mixture Discriminant Analysis
(RMDA) builds a mixture model classifier with noisy class labels.
robcor
computes robust pairwise correlations based on scale estimates,
particularly on
FastQn().
covRobust
provides the
nearest neighbor variance estimation (NNVE) method of Wang and
Raftery (2002).

Clustering (Multivariate)
:
We are
not
considering clusterresistant variance (/standard error)
estimation (aka "sandwich"). Rather e.g. model based
and hierarchical clustering methodology with a particular emphasis
on robustness: Note that
cluster
's
pam()
implementing "partioning around medians" is partly robust (medians
instead of very unrobust kmeans) but is
not
good enough,
as e.g., the k clusters could consist of k1 outliers one
cluster for the bulk of the remaining data.
"Truly" robust clustering is provided by packages
genie,
Gmedian,
otrimle
(trimmed MLE modelbased)
snipEM, (snipping EM)
and
qclust
(robust estim. of Gaussian mixtures) and
notably
tclust
(robust trimmed clustering).
See also the CRAN task views
Multivariate
and
Cluster

Large Data Sets
:
BACON()
(in
robustX)
should be applicable for larger (n,p) than traditional robust
covariance based outlier detectors.
OutlierDM
detects outliers for replicated highthroughput data.
(See also the CRAN task view
MachineLearning.)

Descriptive Statistics / Exploratory Data Analysis
:
boxplot.stats(), etc mentioned above

Time Series
:

R's
runmed()
provides
most robust
running median filtering.

Package
robfilter
contains robust regression and
filtering methods for univariate time series, typically based on
repeated (weighted) median regressions.

The
RobPer
provides several methods for robust
periodogram estimation, notably for irregularly spaced time series.

Peter Ruckdeschel has started to lead an effort for a robust
timeseries package, see
robustts
on RForge.

Further, robKalman,
"Routines for Robust Kalman
Filtering  the ACM and rLSfilter"
, is being developed, see
robkalman
on RForge.
Note however that these (last two items) are not yet available from CRAN.

Econometric Models
:
Econometricians tend to like HAC (heteroscedasticity and
autocorrelation corrected) standard errors. For a broad class of
models, these are provided by package
sandwich.
Note that
vcov(lmrob())
also uses a version of HAC
standard errors for its robustly estimated linear models.
See also the CRAN task view
Econometrics

Robust Methods for Bioinformatics
:
There are several packages in the
Bioconductor project
providing specialized robust methods.
In addition,
RobLoxBioC
provides infinitesimally robust
estimators for preprocessing omics data.

Robust Methods for Survival Analysis
:
Package
coxrobust
provides robust estimation in the Cox
model.
OutlierDC
detects outliers using quantile regression for
censored data.

Robust Methods for Surveys
:
On Rforge only, package
rhte
provides a robust
HorvitzThompson estimator.

Geostatistics
:
Package
georob
aims at robust geostatistical
analysis of spatial data, such as kriging and more.

Collections of several methodologies
:

WRS2
contains
robust tests for ANOVA and ANCOVA and other functionality from
Rand Wilcox's collection.

walrus
builds on
WRS2
's computations,
providing a different user interface.

robeth
contains R functions interfacing to the extensive
RobETH fortran library with many functions for regression,
multivariate estimation and more.

Other approaches to robust and resistant methodology
:

The package
distr
and its several child packages
also allow to explore robust estimation concepts, see e.g.,
distr
on RForge.

Notably, based on these,
the project
robast
aims for the implementation of R
packages for the computation of optimally robust estimators and
tests as well as the necessary infrastructure (mainly S4 classes
and methods) and diagnostics; cf. M. Kohl (2005).
It includes the R packages
RandVar,
RobAStBase,
RobLox,
RobLoxBioC,
RobRex.
Further,
ROptEst, and
ROptRegTS.

RobustAFT
computes Robust Accelerated Failure
Time Regression for Gaussian and logWeibull errors.

wle
Weighted Likelihood Estimation
provides
robustified likelihood estimation for a range of models,
notably (generalized) regression, and time series (AR and
fracdiff).

robumeta
for robust variance metaregression;
metaplus
adds robustness via t or mixtures of
normal distributions.

ssmrob
provides robust estimation and inference in sample selection models.