Re: Correlation with nan package

[Alois Schlögl]

Hi,


currently, corrcoef from the nan-toolbox uses the faster one-pass 
algorithm, that is known to have reduced accuracy.
It's an issue when the data has a very low coefficient of variation
(i.e. std(x)/mean(x) is in the order of sqrt(eps) ) . (for details see
also [1]).

There is a simple workaround, just remove the mean before doing corrcoef.
It will produce the same correlation, and is not subject to the limited
accuracy.
You can do this e.g. with:
    corrcoef(center(a),center(b))

This approach corresponds to the more accurate two-pass algorithm.
Eventually, I might implement the more accurate two-pass algorithm.
Because it has some implications in terms of speed and memory
requirement, users might want to control whether one or the other
algorithm is used.

I hope this answers your questions,

Cheers,
  Alois
 
 


[1]
https://stats.stackexchange.com/questions/94056/instability-of-one-pass-algorithm-for-correlation-coefficient

 


On 2018-07-15 02:01, shivax wrote:
> hi all,
>
> i'm using this package
>
> i use corrcoef(a,b)
>
> but sometimes i see this return: Inf (infinite)
>
>
> how it is possible to return Inf if correlation varies from a minimum of 1
> to -1?
>
>
>
>
>
>
>
>
> --
> Sent from: http://octave.1599824.n4.nabble.com/Octave-General-f1599825.html
>
>