[R] N Sizes between Pairs of Columns using cor(, , , use = 'pairwise')

[William Dunlap]

crossprod(!is.na(tmp))

Bill Dunlap
TIBCO Software
wdunlap tibco.com


On Tue, Jan 21, 2020 at 11:56 AM Doran, Harold <HDoran using air.org> wrote:

> I'm trying to find an efficient way to find the N size on correlations
> produced when using the pairwise option in cor().
>
> Here is a sample to illustrate:
>
> ### Create a sample data frame
> tmp <- data.frame(v1 = rnorm(10), v2 = rnorm(10), v3 = rnorm(10), v4 =
> rnorm(10))
>
> ### Create some random missingness
> for(i in 1:4) tmp[sample(1:10, 2, replace = FALSE), i] <- NA
>
> ### Correlate
> cor(tmp, use = 'pairwise')
>
> Now, a REALLY bad idea would be this (but conceptually it illustrates what
> I want)
>
> ### Identify all column pairs
> pairs <- combn(4,2)
>
> ### Now, write code to loop over each pair of columns and identify where
> both rows are TRUE
> !is.na(tmp[, pairs[,1]])
>
> Of course doing this when the number of pairwise combinations is silly.
> So, hmmm, I don't see as a by-product of the cor() function N sizes, and
> certainly looping over pairs of columns would be doable, but not efficient,
> but any suggestions on this?
>
> Thanks,
> Harold
>
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.
>

	[[alternative HTML version deleted]]