[R] simulate data from multivariate normal with pre-specified correlation matrix
Jiao Yang
yj316 at gwu.edu
Mon Aug 13 23:15:12 CEST 2007
For example, the correlation matrix is 3x3 and looks like
1 0.75 0 0 0
0.75 1 0 0 0
0 0 0 0 0
Can I write the code like this?
p<- 3 # number of variables per observation
N<- 10 # number of samples
# define population correlation matrix sigma
sigma<-matrix(0,p,p) #creates a px p matrix of 0
rank<-2
for (i in 1:rank){
for (j in 1:rank){
rho<-0.75
sigma[i,j]<-rho^abs(i-j)
sigma[i,i]<-1
}
}
# create a data set of 10 observations from multivariate normalcovariance matrix sigma
library(MASS)
Xprime<-mvrnorm(N,rep(1,p), sigma )
Also, if I calculate
S<- cov(Xprime)
will S be the sample correlation matrix?
Thank you very much for your time!
More information about the R-help
mailing list