Replace certain values of a vector with NA

Question

I am trying to replace the values of the vector with NA if the value == -273.17.

Here's a sample data:

x <- c(32,33,34,-273.17,32)

I would like to get this:

x <- c(32,33,34,NA,32)

Here's what I tried

x <- apply(x, 1, 
                (if (x==-273.17)
                 {x==NA} ))

But I get an error instead. Can anyone help me please?

score 3 · Accepted Answer · answered Oct 22 '13 at 07:39

Boolean indexing, similar to answer from @Konrad.Rudolph, using tolerance based on default for ?all.equal:

tol <- 0.5 * .Machine$double.eps # tolerance close to machine precision
float.equal <- function(a, b, p.tol = tol) { abs(a - b) < p.tol }
x[float.equal(x, -273.17)] <- NA

Consider the following if your requirement can be reduced to an easier-to-code predicate, e.g.:

x[x<0] <- NA

all.equal works on vectors of equal length, and reports on the equality of full objects being compared, not element-wise equality.

For R version 3.0.1 (2013-05-16), isTRUE(all.equal()) does not produce the desired result:

> all.equal(x, -273.17)
[1] "Numeric: lengths (5, 1) differ"
> isTRUE(all.equal(x, -273.17))
[1] FALSE

I didn't test Konald's answer do I didn't know it doesn't work anymore. Thanks! — Wet Feet, Oct 22 '13 at 07:45

score 2 · Answer 2 · answered Oct 22 '13 at 07:09

You’d normally use subsetting for this:

x[x == -273.17] <- NA

However, beware of comparing floating point numbers in this fashion. all.equal is a better alternative because it takes into account the inaccuracy in representation:

x[isTRUE(all.equal(x, -273.17))] <- NA

zero323 · Answer 3 · 2013-10-22T07:17:49.743

1

You can use ifelse

ifelse(x==-273.17, NA, x)

but I like @KonradRudolph answer more.

edited Oct 22 '13 at 07:17

answered Oct 22 '13 at 07:07

zero323

322,348
103
959
935

Your's is easier to understand though. – Wet Feet Oct 22 '13 at 07:22

Replace certain values of a vector with NA

3 Answers3

Linked