polyfit: add a third output argument, mu [PATCH]

[Shaun Jackman-5]

The following patch adds a third output argument, mu for compatibility with
Matlab. I made an effort to update the documentation, but I'm not familiar with
Octave's documentation system.

Cheers,
Shaun

2008-06-30  Shaun Jackman  <sjackman@...>

        * polynomial/polyfit.m (polyfit): Add a third output argument,
        mu, for compatibility. If three output arguments are requested,
        the input argument x is shifted and scaled before fitting.

--- /opt/octave-3.0.1/share/octave/3.0.1/m/polynomial/polyfit.m.orig 2008-04-25
09:17:07.000000000 -0700
+++ /opt/octave-3.0.1/share/octave/3.0.1/m/polynomial/polyfit.m 2008-06-30
11:15:46.000000000 -0700
@@ -51,6 +51,10 @@
  ## @item yf
  ## The values of the polynomial for each value of @var{x}.
  ## @end table
+##
+## If three output arguments are requested, the input argument @var{x}
+## is transformed before fitting by @code{(x - mu(1)) / mu(2)},
+## where @code{mu(1) = mean(x)} and @code{mu(2) = std(x)}.
  ## @end deftypefn

  ## Author: KH <Kurt.Hornik@...>
@@ -72,6 +76,12 @@
      error ("polyfit: n must be a nonnegative integer");
    endif

+  if (nargout > 2)
+  mu(1) = mean(x);
+  mu(2) = std(x);
+  x = (x - mu(1)) / mu(2);
+  endif
+
    y_is_row_vector = (rows (y) == 1);

    l = length (x);

[Dmitri A. Sergatskov]

On Mon, Jun 30, 2008 at 1:52 PM, Shaun Jackman <sjackman@...> wrote:
> The following patch adds a third output argument, mu for compatibility with
> Matlab. I made an effort to update the documentation, but I'm not familiar
> with Octave's documentation system.
>

There is wpolifit in octave-forge that has this and more. And I still
do not understand
why it is not in the main octave tree yet...

> Cheers,
> Shaun

Sincerely,

Dmitri.
--

[David Bateman-3]

Dmitri A. Sergatskov wrote:
> On Mon, Jun 30, 2008 at 1:52 PM, Shaun Jackman <sjackman@...> wrote:
>> The following patch adds a third output argument, mu for compatibility with
>> Matlab. I made an effort to update the documentation, but I'm not familiar
>> with Octave's documentation system.
>>
>
> There is wpolifit in octave-forge that has this and more. And I still
> do not understand
> why it is not in the main octave tree yet...


Because there aren't that many of us who are porting the octave-forge
functions to Octave. I did a bit of this but searched for candidates
based on the function name clashes, and as wpolyfit isn't polyfit I
didn't identify this as a replacement.. I'm sure if you go over wpolyfit
 doing the following

* Add GPL copyright (but keep Paul's public domain copyright)
* Change '' quoted strings to "" quoted strings
* Use parenthesis in if statements
* Write "length (varargin)" and not "length(varargin)" etc
* Other standard Octave style fixes

then John would accept this as a replacement if its clearly better.

D.

[Ben Abbott]

On Jul 1, 2008, at 2:23 AM, David Bateman wrote:

> Dmitri A. Sergatskov wrote:
>> On Mon, Jun 30, 2008 at 1:52 PM, Shaun Jackman <sjackman@...>  
>> wrote:
>>> The following patch adds a third output argument, mu for  
>>> compatibility with
>>> Matlab. I made an effort to update the documentation, but I'm not  
>>> familiar
>>> with Octave's documentation system.
>>>
>>
>> There is wpolifit in octave-forge that has this and more. And I still
>> do not understand
>> why it is not in the main octave tree yet...
>
>
> Because there aren't that many of us who are porting the octave-forge
> functions to Octave. I did a bit of this but searched for candidates
> based on the function name clashes, and as wpolyfit isn't polyfit I
> didn't identify this as a replacement.. I'm sure if you go over  
> wpolyfit
> doing the following
>
> * Add GPL copyright (but keep Paul's public domain copyright)
> * Change '' quoted strings to "" quoted strings
> * Use parenthesis in if statements
> * Write "length (varargin)" and not "length(varargin)" etc
> * Other standard Octave style fixes
>
> then John would accept this as a replacement if its clearly better.
>
> D.

Hey guys,

A changeset was applied back in Feb that includes the third output  
argument, mu.

        http://www.nabble.com/Re%3A--changeset--Polyfit-with-scaling-p15834853.html

I checked my copy of jwe's default archive, and it is present there.  
Is it not in yours?

Ben

[David Bateman]

Ben Abbott wrote:

> Hey guys,
>
> A changeset was applied back in Feb that includes the third output
> argument, mu.
>
>     http://www.nabble.com/Re%3A--changeset--Polyfit-with-scaling-p15834853.html 
>
>
> I checked my copy of jwe's default archive, and it is present there.
> Is it not in yours?

Ok, so what else is in wpolyfit.. It is always better to not have two
versions of the same function, as thats just a source of confusion, so
it would be better to cut any beneficial code from wpolyfit and port it
to Octave.

D.

[Shaun Jackman-5]

Ben Abbott wrote:
> On Jul 1, 2008, at 2:23 AM, David Bateman wrote:
>> Dmitri A. Sergatskov wrote:
>>> On Mon, Jun 30, 2008 at 1:52 PM, Shaun Jackman <sjackman@...>
>>> wrote:
>>>> The following patch adds a third output argument, mu for
>>>> compatibility with
>>>> Matlab. I made an effort to update the documentation, but I'm not
>>>> familiar
>>>> with Octave's documentation system.
...
> A changeset was applied back in Feb that includes the third output
> argument, mu.
>
>     http://www.nabble.com/Re%3A--changeset--Polyfit-with-scaling-p15834853.html 
>
>
> I checked my copy of jwe's default archive, and it is present there. Is
> it not in yours?

I'm running Octave 3.0.1, which does not include this patch. Any upcoming plans
for a stable release?

Cheers,
Shaun

[John W. Eaton]

On  2-Jul-2008, Shaun Jackman wrote:

| I'm running Octave 3.0.1, which does not include this patch. Any upcoming plans
| for a stable release?

Is there anyone out there who would like to become the release manager
for the stable release series?

jwe