On 29/09/2009, at 10:54 PM, Steve Jones wrote:
Hi List,
I'm trying to perform some spectral analysis on a time series, but
I've
got a few data points missing, so they have NA values. Unfortunately,
the fft function doesn't seem to like this, and gives a completely
empty
result.
Other than making up values to fill in the gaps, is there any way I
can
get around this?
Thanks in advance,
Steve.
PS I haven't spotted anything relevant in the man page, and as fft
is an
internal function, I'll have to get the source code to figure out
what's
going on. I'll do it if I must, but I'd rather not!
I'm curious. Just what do you think the discrete Fourier transform of
a sequence of values containing missing values, should be? Remember
that
the fast Fourier transform is simply an algorithm for the efficient
*computation* of the discrete Fourier transform.
You can very easily write your own ``sft()'' (slow Fourier
transform :-) )
function to calculate the DFT directly from the definition. Unless your
data sequences are incredibly large, the speed difference between fft
() and
sft() will not be noticeable. Into your sft() function you can place
whatever
technique you think appropriate for handling missing values. It is
not at all
clear to me that anything appropriate is possible.
cheers,
Rolf Turner
##
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}
__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
