[R] getting covariance ignoring NaN missing values

[arun]

Hi,
If you look at ?cov(),
there are options for 'use':
set.seed(15)
a=array(rnorm(9),dim=c(3,3))
 a[3,2]<- NaN

 cov(a,use="complete.obs")
#           [,1]        [,2]       [,3]
#[1,]  1.2360602 -0.32167789  0.8395953
#[2,] -0.3216779  0.08371491 -0.2185001
#[3,]  0.8395953 -0.21850006  0.5702960
 cov(a,use="na.or.complete")
#           [,1]        [,2]       [,3]
#[1,]  1.2360602 -0.32167789  0.8395953
#[2,] -0.3216779  0.08371491 -0.2185001
#[3,]  0.8395953 -0.21850006  0.5702960
 cov(a,use="pairwise.complete.obs")
#           [,1]        [,2]       [,3]
#[1,]  1.2570603 -0.32167789  0.7377472
#[2,] -0.3216779  0.08371491 -0.2185001
#[3,]  0.7377472 -0.21850006  0.4433438
A.K.




----- Original Message -----
From: Sachinthaka Abeywardana <sachin.abeywardana at gmail.com>
To: "r-help at r-project.org" <r-help at r-project.org>
Cc: 
Sent: Thursday, March 7, 2013 10:36 PM
Subject: [R] getting covariance ignoring NaN missing values

Hi all,

I have a matrix that has many NaN values. As soon as one of the columns has
a missing (NaN) value the covariance estimation gets thrown off.

Is there a robust way to do this?

Thanks,
Sachin

a=array(rnorm(9),dim=c(3,3))> a            [,1]       [,2]      [,3]
[1,] -0.79418236  0.7813952  0.855881
[2,] -1.65347906 -1.9462446 -0.376325
[3,] -0.03144987  0.6756862 -1.879801> a[3,2]=NANError: object 'NAN'
not found> a[3,2]=NaN> a            [,1]       [,2]      [,3]
[1,] -0.79418236  0.7813952  0.855881
[2,] -1.65347906 -1.9462446 -0.376325
[3,] -0.03144987        NaN -1.879801> cov(a)           [,1] [,2]       [,3]
[1,]  0.6585217   NA -0.5777408
[2,]         NA   NA         NA
[3,] -0.5777408   NA  1.8771214

    [[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.