R ts object - Preserve index while removing NAs

Question

Consider a time series object with several NA values:

x <- seq(10)
x[seq(2,10,2)] <- NA
x <- ts(x)

Here's the default index:

index(x)
[1]  1  2  3  4  5  6  7  8  9 10

If I remove the NA values using na.exclude, I get a new index:

na.exclude(x)
[1] 1 3 5 7 9
attr(,"na.action")
[1]  2  4  6  8 10
attr(,"class")
[1] "exclude"    
index(na.exclude(x))
[1] 1 2 3 4 5

If I try na.omit I just get an error:

na.omit(x)
Error in na.omit.ts(x) : time series contains internal NAs
> index(na.omit(x))
Error in na.omit.ts(x) : time series contains internal NAs

If I try setting the index to the value I need (and which I expect should be the natural result of removing NAs), I get another error:

index(x) <- c(1,3,5,7,9)
Error in UseMethod("index<-") : no applicable method for 'index<-' 
applied to an object of class "ts"

Is there a direct way to remove the NA values while preserving the original index for the non-NAs? I would prefer not to use xts, zoo, etc., since there are some functions in the package stats that only accept ts objects.

Thanks.

Answer 1

You could do this with na.remove from the tseries package (which will also keep the ts class):

x <- seq(10)
x[seq(2,10,2)] <- NA
x <- ts(x)

And then:

library(tseries)
na.remove(x)
#Time Series:
#Start = 1 
#End = 9 
#Frequency = 0.5 
#[1] 1 3 5 7 9
#attr(,"na.removed")
#[1]  2  4  6  8 10

index(na.remove(x))
#[1] 1 3 5 7 9

The ts class remains as is:

class(x)
#[1] "ts"
class(na.remove(x))
#[1] "ts"