On Thu, Oct 15, 2020 at 10:41 PM ilja <ilibas@uns.ac.rs> 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