Skip to content

generate fourth vector based on known correlations

2 messages · bkelcey at umich.edu, Greg Snow

Original
bkelcey at umich.edu
# 
I am trying to generate a fourth vector,z, given three known and fixed 
vectors, x1,x2,x3 with corresponding known and fixed correlations with 
themeselves and with z. That is, all correlations are known and 
prespecified. How can I do this?
Thank you,
ben
Greg Snow
# 
Here is one approach (someone with better linear algebra skills may be
able to shorten this, some steps could be combined, but I wanted to show
each step):


# create x1-x3

x1 <- rnorm(100, 50, 3)
x2 <- rnorm(100) + x1/5
x3 <- rnorm(100) + x2/5

# find current correlations

cor1 <- cor( cbind(x1,x2,x3) )
cor1

# create 1st version of z

z <- rnorm(100)

# combine in a matrix

m1 <- cbind( x1, x2, x3, z )

# center and scale

m2 <- scale(m1)

# find cholesky decomp

c1 <- chol(var(m2))

# force to be independent

m3 <- m2 %*% solve(c1)

zapsmall(cor(m3)) # check

# create new correlation matrix:

cor2 <- cbind( rbind( cor1, z=c(.5,.3,.1) ), z=c(.5,.3,.1,1) )

# create new matrix

m4 <- m3 %*% chol(cor2)

# uncenter and unscale

m5 <- sweep( m4, 2, attr(m2, 'scaled:scale'), '*')
m5 <- sweep( m5, 2, attr(m2, 'scaled:center'), '+')

# check to see if we succeeded

all.equal( cbind(x1,x2,x3), m5[, 1:3] )
all.equal( cor(m5)[1:3,1:3], cor1 )
all.equal( cor(m5), cor2 )

cor(m5)

 
Hope this helps,