Question regarding to replace <NA>

On Tue, Nov 9, 2010 at 9:39 AM, Joshua Wiley <jwiley.psych at gmail.com> wrote:

Hi Kate,

is.na() does not work on entire data frames.

whoops, I did not mean that. ?is.na() has a data frame method, but
there are assignment issues (as you saw) when you use it that way.

Last correction (I promise).  Somehow I tested incorrectly which lead
to me to my two erroneous statements before.  is.na() works just fine
with data frames.  It was just the factor issue (as Phil said).

x <- data.frame(a = 1:3, b = c(1, 2, NA), d = c(NA, 2, 3), e = factor(c(NA, NA, "1")))
x

a  b  d    e
1 1  1 NA <NA>
2 2  2  2 <NA>
3 3 NA  3    1

x[is.na(x)] <- 0

Warning message:
In `[<-.factor`(`*tmp*`, thisvar, value = 0) :
  invalid factor level, NAs generated

x # other NAs removed, but not the factor ones

a b d    e
1 1 1 0 <NA>
2 2 2 2 <NA>
3 3 0 3    1

off-to-hide-under-a-rock-Josh