simcor.H: Simulating the Hub Matrix (entries filled in using Toeplitz...

In richardbeare/simcor: Generate noisy correlation matrices

Description

Simulating the Hub Matrix (entries filled in using Toeplitz structure)

Usage

simcor.H(k = 6, size = c(10, 5, 8, 7, 15, 50), rho = rbind(c(0.9, 0.7),
  c(0.7, 0.7), c(0.7, 0.2), c(0.5, 0.3), c(0.9, 0.85), c(0.3, 0.2)),
  power = 1, epsilon = 0.08, eidim = 2)

Arguments

k	is the number of groups
size	is a vector of length k specifying the size of each group
rho	is a vector of length k specifying base correlation values
power	= 2 makes the correlations stay high, = 0.5 makes the correlations descent rapidly
epsilon	<- (1-min(rho) - 0.75*min(tau) ) - .01 - maximum entry-wise random noise.
eidim	is the space from which the noise is generated, the smaller the more noise

Examples

# Figure 2 in the paper
vw <- function(Im)
{
image(Im[, ncol(Im):1], zlim=c(-0.4, 1), col=rev(heat.colors(64)))
}
k<-3
sz <- c(100, 50, 80)
rho <- matrix(c(0.7, 0.7, 0.4, 0, 0, 0), nrow=3)
epsilon <- 0.23
hTC1 <- simcor.H(k=k, size=sz, rho=rho, power=1, epsilon=0.23, eidim=2)
vw(hTC1)
rho <- matrix(c(0.7, 0.7, 0.4, 0.5, 0.6, 0.2), nrow=3)

hTC2 <- simcor.H(k=k, size=sz, rho=rho, power=1, epsilon=0.29, eidim=2)
vw(hTC2)

hTC3 <- simcor.H(k=k, size=sz, rho=rho, power=1, epsilon=0.29, eidim=25)
vw(hTC3)

hTC4 <- simcor.H(k=k, size=sz, rho=rho, power=1, epsilon=0.1, eidim=2)
vw(hTC4)
hTC5 <- simcor.H(k=k, size=sz, rho=rho, power=1, epsilon=0.25, eidim=2)
vw(hTC5)
rho <- matrix(c(0.8, 0.75, 0.7, 0, 0, 0), nrow=3)
hTC6 <- simcor.H(k=k, size=sz, rho=rho, power=1, epsilon=0.19, eidim=2)
vw(hTC6)

richardbeare/simcor documentation built on May 19, 2019, 4:16 p.m.