NA values not being excluded in `cor`

Question

To simplify, I have a data set which is as follows:

b <- 1:6
# > b
# [1] 1 2 3 4 5 6
jnk <- c(2, 4, 5, NA, 7, 9)
# > jnk
# [1]  2  4  5 NA  7  9

When I try:

cor(b, jnk, na.rm=TRUE)

I get:

> cor(b, jnk, na.rm=T)
  Error in cor(b, jnk, na.rm = T) : unused argument (na.rm = T)

I've also tried na.action = na.exclude, etc. None seem to work. It'd be really helpful to know what the issue is and how I can fix it. Thanks.

score 60 · Answer 1 · edited Aug 04 '18 at 15:57

60

TL; DR: Use instead:

cor(b, jnk, use="complete.obs")

Read ?cor:

cor(x, y = NULL, use = "everything",
     method = c("pearson", "kendall", "spearman"))

It doesn't have na.rm, it has use.

an optional character string giving a method for computing covariances in the presence of missing values. This must be (an abbreviation of) one of the strings "everything", "all.obs", "complete.obs", "na.or.complete", or "pairwise.complete.obs".

Pick one. Details of what each does is in the Details section of ?cor.

edited Aug 04 '18 at 15:57

Michael Ohlrogge

9,131
3
41
69

answered Jul 14 '15 at 16:42

Spacedman

86,225
12
117
197

4

A very annoying part about `R` is the inconsistency w.r.t. `na.rm` and its variants... – MichaelChirico Jul 14 '15 at 18:02
5

at least `help(foo)` is consistent. – Spacedman Jul 14 '15 at 19:13
Yes. Still I hope they (in all `base`/out of the box functions, at least) unify at some point... – MichaelChirico Jul 14 '15 at 20:06
Brilliant thank you! For anyone else looking for the qiock solution, to exclude NA's from the cor() function, use the argument: use="complete.obs" use="complete.obs" – Charlie Jul 15 '15 at 09:53

score 6 · Answer 2 · answered Oct 31 '17 at 02:13

6

Just to make sure the answer to this question is clear.

To ignore NA, use

b <- 1:6
jnk <- c(2, 4, 5, NA, 7, 9)
cor(b, jnk, use="complete.obs")
[1] 0.9905977

answered Oct 31 '17 at 02:13

Steve Lihn

185
2
3

NA values not being excluded in `cor`

2 Answers2

Linked