PCA in Octave

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

PCA in Octave

ilja

In the literature, there are several versions of Principal Components
Analysis mentioned.
Conventional, probabilistic, kernel.
Which of them is implemented in Octave ?

Regards
Ilja



--
Sent from: https://octave.1599824.n4.nabble.com/Octave-General-f1599825.html


Reply | Threaded
Open this post in threaded view
|

Re: PCA in Octave

siko1056
On 10/16/20 5:41 AM, ilja wrote:
>
> In the literature, there are several versions of Principal Components
> Analysis mentioned.
> Conventional, probabilistic, kernel.
> Which of them is implemented in Octave ?
>
> Regards
> Ilja
>

Please take a look at the tisean package [1,2].  This package is bundled
with the MS Windows installer.  Activate it by typing

   pkg load tisean

in the Octave command window.

HTH,
Kai

[1] https://octave.sourceforge.io/tisean/function/pca.html
[2] https://wiki.octave.org/TISEAN_package


Reply | Threaded
Open this post in threaded view
|

Re: PCA in Octave

Andreas Weber-6
Am 16.10.20 um 10:56 schrieb Kai Torben Ohlhus:
> Please take a look at the tisean package [1,2].

I've seen that the link to the project
https://bitbucket.org/josiah425/tisean

which is used on various places is broken. Do you know what happened?
-- Andy


Reply | Threaded
Open this post in threaded view
|

Re: PCA in Octave [OF tisean package]

siko1056
On 10/16/20 7:51 PM, Andreas Weber wrote:
> Am 16.10.20 um 10:56 schrieb Kai Torben Ohlhus:
>> Please take a look at the tisean package [1,2].
>
> I've seen that the link to the project
> https://bitbucket.org/josiah425/tisean
>
> which is used on various places is broken. Do you know what happened?
> -- Andy
>

OF has a backup of the repo [3].

Let's ask the developer: Piotr Held can you clarify and fix the proper
link to your active development repository?

Kai

[3] https://sourceforge.net/p/octave/tisean/ci/default/tree/


Reply | Threaded
Open this post in threaded view
|

Re: PCA in Octave [OF tisean package]

Juan Pablo Carbajal-2
On Fri, Oct 16, 2020 at 1:27 PM Kai Torben Ohlhus <[hidden email]> wrote:

>
> On 10/16/20 7:51 PM, Andreas Weber wrote:
> > Am 16.10.20 um 10:56 schrieb Kai Torben Ohlhus:
> >> Please take a look at the tisean package [1,2].
> >
> > I've seen that the link to the project
> > https://bitbucket.org/josiah425/tisean
> >
> > which is used on various places is broken. Do you know what happened?
> > -- Andy
> >
>
> OF has a backup of the repo [3].
>
> Let's ask the developer: Piotr Held can you clarify and fix the proper
> link to your active development repository?
>
> Kai
>
> [3] https://sourceforge.net/p/octave/tisean/ci/default/tree/
>
I do not think Piotr is still around; and bitbucket has killed all hg repos.
That link was to his development version, the OF package should use in
any official link.

Beware that for pca we have open issues, e.g.
https://savannah.gnu.org/bugs/?56698


Reply | Threaded
Open this post in threaded view
|

Re: PCA in Octave

Juan Pablo Carbajal-2
In reply to this post by ilja
On Thu, Oct 15, 2020 at 10:41 PM ilja <[hidden email]> wrote:
>
>
> In the literature, there are several versions of Principal Components
> Analysis mentioned.
> Conventional, probabilistic, kernel.
> Which of them is implemented in Octave ?
>
> Regards
> Ilja

Most likely you will find the conventional and probabilistic from
(they differ on the normalization only).
These flavors are just the massaged output of SVD [1]. I have some
very incomplete notes on this [2], as well, use carefully.

For kernel pca, you can check the web, e.g. https://github.com/steven2358/kmbox.
If you are using PCA for system reduction check the empirical gramian
package: https://gramian.de/

[1]: https://stats.stackexchange.com/questions/134282/relationship-between-svd-and-pca-how-to-use-svd-to-perform-pca
[2]: https://gitlab.com/kakila/PCAR_toolbox/-/blob/master/doc/SVD_Notes.tex


Reply | Threaded
Open this post in threaded view
|

Re: PCA in Octave

ilja
Juan muchas gracias.

Ilja

On 20/10/2020 08:49, Juan Pablo Carbajal wrote:

> On Thu, Oct 15, 2020 at 10:41 PM ilja <[hidden email]> wrote:
>>
>> In the literature, there are several versions of Principal Components
>> Analysis mentioned.
>> Conventional, probabilistic, kernel.
>> Which of them is implemented in Octave ?
>>
>> Regards
>> Ilja
> Most likely you will find the conventional and probabilistic from
> (they differ on the normalization only).
> These flavors are just the massaged output of SVD [1]. I have some
> very incomplete notes on this [2], as well, use carefully.
>
> For kernel pca, you can check the web, e.g. https://github.com/steven2358/kmbox.
> If you are using PCA for system reduction check the empirical gramian
> package: https://gramian.de/
>
> [1]: https://stats.stackexchange.com/questions/134282/relationship-between-svd-and-pca-how-to-use-svd-to-perform-pca
> [2]: https://gitlab.com/kakila/PCAR_toolbox/-/blob/master/doc/SVD_Notes.tex

--
Pozdrav
prof. Ilija Bašičević, phd

http://member.acm.org/~ilija.basicevic