a matrix or data frame. Columns represent variables, and rows represent
observations.
OPTIONAL ARGUMENTS:
corr
logical flag; if
corr=T the estimated
correlation matrix is returned.
center
logical flag or numeric vector containing the location about which the
covariance is to be taken. If
center=T then a
robust estimate of the center is computed; if
center=F then no centering takes place and the
center is set equal to zero. The center argument is only used by the
Stahel-Donoho estimator (which is not the default estimator).
distance
logical flag; if
distance=T the Mahalanobis
distances are computed.
na.action
a function to filter missing data. The default (with na.fail) is to create
an error if any missing values are found. A possible alternative is
na.omit, which deletes observations that contain one or more missing
values.
estim
the robust estimator used by covRob. The choices are: "mcd" for the Fast
MCD algorithm of Rousseeuw and Van Driessen, "donostah" for the
Donoho-Stahel projection based estimator, "M" for the constrained M
estimator provided by Rocke, "pairwiseQC" for the quadrant correlation
based pairwise estimator, and "pairwiseGK" for the Gnanadesikan-Kettenring
pairwise estimator. The default "auto" selects from "donostah", "mcd", and
"pairwiseQC" with the goal of producing a good estimate in a resonable
amount of time.
control
a list of control parameters to be used in the numerical algorithms.
See
for the possible control parameters and their default settings.
...
control parameters in
may be passed
directly to
covRob.
VALUE:
an object of class
"covRob" with components:
cov
the robust estimate of the covariance/correlation matrix.
center
a robust estimate (or the specification) of the location vector.
dist
the Mahalanobis distances computed with a robust estimate of covariance,
only returned if
distance=T.
iter
the number of iterations used.
evals
eigenvalues of the covariance matrix if
corr=F
; otherwise eigenvalues of the
correlation matrix.
NOTE:
Almost all of the estimation methods require that
the number of rows of data be more than twice the
number of columns. The pairwise estimation methods
don't strictly require this, but they often fail
on short datasets. You may want to use something
like
if(is( cov <- try(covRob(x)), "Error")) cov <- list(cov=var(x))
to get a robust estimate of covariance if it works and fall back
to a classical estimate otherwise.
REFERENCES:
Marazzi, A. (1993).
Algorithms, routines, and S functions for robust statistics.
Wadsworth & Brooks/Cole, Pacific Grove, CA.
Rousseeuw, P.J. and Van Driessen, K. (1999). A Fast Algorithm for the
Minimum Covariance Determinant Estimator,
Technometrics,
41, 212-223.
SEE ALSO:
,
.
EXAMPLES:
covRob(stack.dat)
covRob(stack.dat, estim = "mcd", quan = .75, ntrial = 1000)