On Thu, May 31, 2012 at 4:42 PM, Juan Pablo Carbajal
<carba...@ifi.uzh.ch> wrote:
> Hi,
>
> I have a question concerning xcorr for the case of vector input.
>
> The results of xcorr do not match with the ones suing the straight formula
> http://en.wikipedia.org/wiki/Autocorrelation#Estimation
>
> In particular taking the vector
> x = rand(4069,1);
>
> xcorr with option 'coeff' and 'biased' show a linear trend that the
> mentioned formula doesn't show. I guess this comes from the fact that
> the data has nonzero mean value and the formula removes it, while
> xcorr doesn't.
>
> Is this the expected behavior?

Yes.  In fact rewind a little bit on the page you cite:
http://en.wikipedia.org/wiki/Autocorrelation#Signal_processing

    In signal processing, the above definition is often used without the
    normalization, that is, without subtracting the mean and dividing by
    the variance. When the autocorrelation function is normalized by
    mean and variance, it is sometimes referred to as the
    autocorrelation coefficient.

"Normalization" here is referring to the data samples, as opposed to
the "scale" options that xcorr accepts.

> If it is, we should improve the help text to mention this fact explicitly.

I think this is typical of different interpretations of some
terminology between statistics and signal processing.  If you want to
propose some clarifying text, keep in mind that for signal processing
oriented users this is naturally the correct behavior.  See for
example "help autocov" in Octave core.

-- 
mike

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Octave-dev mailing list
Octave-dev@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/octave-dev

Reply via email to