corpcor (version 1.3.0)

cor2pcor: Compute Partial Correlation from Correlation Matrix -- and Vice Versa

Description

cor2pcor computes the pairwise partial correlation coefficients from either a correlation or a covariance matrix.

pcor2cor takes either a partial correlation matrix or a partial covariance matrix as input, and computes from it the corresponding correlation matrix. cov2pcov transform a covariance matrix into a partial covariance matrix, and pcov2cov performs the corresponding backtransform.

Usage

cor2pcor(m, tol)
pcor2cor(m, tol)
cov2pcov(m, tol)
pcov2cov(m, tol)

Arguments

m

covariance matrix or (partial) correlation matrix

tol

tolerance - singular values larger than tol are considered non-zero (default value: tol = max(dim(m))*max(D)*.Machine$double.eps). This parameter is needed for the singular value decomposition on which

Value

• A matrix with the pairwise partial correlation coefficients (cor2pcor), with pairwise correlations (pcor2cor), with the partial covariances (cov2pcor), or the covariances (pcov2cov).

Details

The partial covariance is the inverse of the covariance matrix, with the signs of the off-diagonal elements reversed. The partial correlation is the standardized partial covariance matrix.

In graphical Gaussian models the partial correlations represent the direct interactions between two variables, conditioned on all remaining variables. In all the above functions the pseudoinverse is employed for inversion - hence even singular covariances (with some zero eigenvalues) may be used. However, a better option may be to estimate a positive definite covariance matrix using cov.shrink. Note that for efficient computation of partial correlation coefficients from data x it is recommend to employ pcor.shrink(x), and *not* cor2pcor(cor.shrink(x)).

References

Whittaker J. (1990). Graphical Models in Applied Multivariate Statistics. John Wiley, Chichester.

See Also

pcor.shrink, pseudoinverse.

Examples

# load corpcor library
library("corpcor")

# covariance matrix
m.cov <- rbind(
 c(3,1,1,0),
 c(1,3,0,1),
 c(1,0,2,0),
 c(0,1,0,2)
)
m.cov


# partial covariance matrix
m.pcov <- cov2pcov(m.cov)
m.pcov


# corresponding correlation matrix
m.cor.1 <- cov2cor(m.cov)
m.cor.1

# compute partial correlations (from covariance matrix)
m.pcor.1 <- cor2pcor(m.cov)
m.pcor.1

# compute partial correlations (from correlation matrix)
m.pcor.2 <- cor2pcor(m.cor.1)
m.pcor.2

# compute partial correlations (from partial covariance matrix)
m.pcor.3 <- cov2cor(m.pcov)
m.pcor.3


zapsmall( m.pcor.1 ) == zapsmall( m.pcor.2 )

# backtransformation
m.cor.2 <- pcor2cor(m.pcor.1)
m.cor.2
zapsmall( m.cor.1 ) == zapsmall( m.cor.2 )