Recommendation for Octave statistics functions

[Ben Abbott]

On Jan 24, 2008, at 3:39 AM, John W. Eaton wrote:

> On 21-Jan-2008, Michael D. Godfrey wrote:
>
> | There has been some discussion of some of the statistics
> | functions in Octave such as median and discrete_inv,
> | and related functions from Octave-forge.  This has often
> | involved compatibility issues.  And, not infrequently, there
> | is a choice between compatibility with Matlab, S, R, or
> | Mathematica.
> |
> | It would be good to decide which of the alternatives deserves
> | priority for compatibility.  My preference would be R, S,  
> Mathematica,
> | then Matlab.  R and S are likely to be the same, but since the S
> | author now works on R, R should be given some preference.
> | In addition, since R is under some GPL it is easier to check
> | for consistency.
>
> One of the goals of Octave 3.1 is to remove some of the non-core
> functions from Octave, including most of the statistics functions.
> Since this is not my area, I'd be happy to have someone else take over
> maintenance of these functions once they are extracted from Octave
> into a separate package.  In that case, it would be up to the
> maintainer of the package whether the package is compatible with
> Matlab.  My guess is that users would want that, and that it would be
> bad to have functions with the same signature behave differently in
> Matlab and Octave.
>
> jwe


How will the determination be made as to what are core functions and  
non-core functions?

Ben