Efficient Estimation of Covariance and (Partial) Correlation
Description
This package implements a shrinkage estimator to allow
the efficient inference of large-scale covariance matrices
from small sample data. The resulting estimates are always
positive definite, much more accurate than the empirical estimate,
well conditioned, computationally inexpensive, and require
only little a priori modeling. The package also contains
similar functions for inferring variances, correlations and partial
correlations and partial covariances. In addition, it provides
functions for fast svd computation, for computing the pseuoinverse,
for checking the rank and positive definiteness of a matrix, and
for the computationally fast inversion of the covariance
and correlation matrix.