[R] acf and na.pass

[Stéphane Dray]

Hello,
I would like to have some information about acf with missing values.
Let us consider this example:
x=rnorm(100)
x2=x
x2[sample(100,10)]=NA
acf1=acf(x)
acf2=acf(x2,na.action=na.pass)

The computation of the acf is different for the two data sets. Looking 
at the the code reveals that with missing values, the computation of acf 
takes into account the number of pairs of non-NA values (i.e. the number 
of pairs of data that are used to compute the acf, and this number is 
different for each lag). This makes sense for me. Where I have more 
problems, concern the computation of confidence intervall. Here, the 
plot.acf function compute the values using x$n.used, which is equal for 
these two cases to 100. Is there some explanation for this choice ?  I 
would think that the computation of CI would also take into account the 
number of pairs of non-NA values. Is there some references on the topic ?

Thanks in advance,
Sincerely.


-- 
Stéphane DRAY (dray at biomserv.univ-lyon1.fr )
Laboratoire BBE-CNRS-UMR-5558, Univ. C. Bernard - Lyon I
43, Bd du 11 Novembre 1918, 69622 Villeurbanne Cedex, France
Tel: 33 4 72 43 27 57       Fax: 33 4 72 43 13 88
http://biomserv.univ-lyon1.fr/~dray/