Error in cor(x[a], x[b], use = "complete.obs") : 'x' must be numeric
This is strange, it works on your example (and you've understood what I'm
trying to do perfectly), but then when I use it on the original data it
comes up with the error above
I've checked str() and the columns are all numeric
???
-----Original Message----- From: Joshua Wiley
Sent: Tuesday, November 15, 2011 7:14 PM
To: robgriffin247
Cc: r-help at r-project.org
Subject: Re: [R] correlations between columns for each row
Hi Rob,
Here is one approach:
## define a function that does the calculations
## (the covariance of two vectors divided by the square root of
## the products of their variances is just a correlation)
rF <- function(x, a, b) cor(x[a], x[b], use = "complete.obs")
set.seed(1)
bigdata <- matrix(rnorm(271 * 13890), ncol = 271)
results <- apply(bigdata, 1, FUN = rF, a = 174:213, b = 214:253)
## combine
bigdata <- cbind(bigdata, iecorr = results)
Hope this helps,
Josh
On Tue, Nov 15, 2011 at 8:42 AM, robgriffin247
<robgriffin247 at hotmail.com> wrote:
Just as an update on this problem:
I have managed to get the variance for the selected columns
Now all I need is the covariance between these 2 selections -
the two target columns are and the aim is that a new column contain a
covariance value between these on each row:
maindata[,c(174:213)] and maindata[,c(214:253]
I've played around with all sorts of apply (and derivatives of apply) and
in
various different setups so I think I'm close but I feel like I'm chasing
my
tail here!
--
View this message in context:
http://r.789695.n4.nabble.com/correlations-between-columns-for-each-row-tp4039193p4073208.html
Sent from the R help mailing list archive at Nabble.com.