cov.mvv: Minimum Variance Vector Covariance Estimator

Description Usage Arguments Value References

Description

This implements the minimum variance vector covariance estimator described in Herwindiati, Djauhari, and Mashuri (2007). It is exactly analagous to the minimum covariance determinant (MCD), except the trace of the matrix is the objective function, rather than the determinant. The deterministic MCD algorithm described in Hubert, Rousseeuw, and Verdonck (2012) is adapted here, rather than adapting the fast MCD algorithm in the paper introducing the MVV.

Usage

1
2
3
4
5
6
cov.mvv(
  x,
  kappa = 0.75,
  method = c("tau", "pb", "bisq", "huber", "mopt", "Qn", "Sn"),
  opts = list(b = 0.1, eff = 0.9)
)

Arguments

x

a data frame or matrix of numeric covariates

kappa

the the proportion of the data to use in each subset. defaults to 0.75. must be > 0.50.

method

the method of measuring location and scale. see cov.mrcd for details on the available options.

opts

list of options for the various scale estimators. "b" determines the percentage bend coefficient for "pb", "eff" determines the efficiency of the "huber", "bisquare" and "mopt" scale estimators.

Value

a covRobust object containing the following elements:

References

Herwindiati, D. E.; Djauhari, M. A.; Mashuri, M. (2007). Robust Multivariate Outlier Labeling. Communications in Statistics - Simulation and Computation, 36(6), 1287–1294. doi:10.1080/03610910701569044

Hubert, M.; Rousseeuw, P.J.; Verdonck, T. (2012) A Deterministic Algorithm for Robust Location and Scatter, Journal of Computational and Graphical Statistics, 21(3), 618-637, doi: 10.1080/10618600.2012.672100


abnormally-distributed/cvreg documentation built on May 3, 2020, 3:45 p.m.