RobStatTM (version 1.0.2)

covRobRocke: Rocke's robust multivariate location and scatter estimator

Description

This function computes Rocke's robust estimator for multivariate location and scatter.

Usage

covRobRocke(X, initial = "K", maxsteps = 5, propmin = 2, qs = 2,
  maxit = 50, tol = 1e-04, cor = FALSE)

Arguments

X

a data matrix with observations in rows.

initial

A character indicating the initial estimator. Valid options are 'K' (default) for the Pena-Prieto 'KSD' estimate, and 'mve' for the Minimum Volume Ellipsoid.

maxsteps

Maximum number of steps for the line search section of the algorithm.

propmin

Regulates the proportion of weights computed from the initial estimator that will be different from zero. The number of observations with initial non-zero weights will be at least p (the number of columns of X) times propmin.

qs

Tuning paramater for Rocke's loss functions.

maxit

Maximum number of iterations.

tol

Tolerance to decide converngence.

cor

A logical value. If TRUE a correlation matrix is included in the element cor of the returned object. Defaults to FALSE.

Value

A list with class “covRob” containing the following elements:

mu

The location estimate

V

The scatter (or correlation) matrix estimate, scaled for consistency at the normal distribution

center

The location estimate. Same as mu above.

cov

The scatter matrix estimate, scaled for consistency at the normal distribution. Same as V above.

cor

The correlation matrix estimate, if the argument cor equals TRUE. Otherwise it is set to NULL.

dista

Robust Mahalanobis distances.

w

weights

gamma

Final value of the constant gamma that regulates the efficiency.

Details

This function computes Rocke's robust estimator for multivariate location and scatter.

References

http://www.wiley.com/go/maronna/robust

Examples

# NOT RUN {
data(bus)
X0 <- as.matrix(bus)
X1 <- X0[,-9]
tmp <- covRobRocke(X1)
round(tmp$cov[1:10, 1:10], 3)
tmp$mu

# }