[R] Sample covariance matrix in R

Ben Bolker bbolker at gmail.com
Thu Nov 18 22:57:45 CET 2010


Alaios <alaios <at> yahoo.com> writes:

> 
> Hello everyone.
> I would like to find the sample covariance matrix using R.
> 
> So far I read on the wikipedia what a sample_covariance is
> http://en.wikipedia.org/wiki/Sample_covariance
> 
> according to wikipedia one vector is 
> enough to calculate the sample covariance matrix.

  I'm sorry, where does it say that??  The expression given
for the sample covariance is

    q_{ij}=\frac{1}{N-1}\sum_{k=1}^{N}
    \left( x_{ik}-\bar{x}_i \right) \left( x_{jk}-\bar{x}_j \right) . 

if N=1 this whole thing will go up in smoke: the denominator and
numerator will both be zero.


> In R I tried cov(myvector) and I get the reply that I need to
>  pass either two argument or one matrix with x,y
> values .
> 
> How can I find the sample covariance matrix?
> 

  You need more than one sample (!); even trying to do it with
two samples would give you a mathematically well-defined but
statistically awful estimate.

  I strongly suggest that you consult a statistics book or
a local expert ...

 good luck



More information about the R-help mailing list