Calculates the pca for data
[E, D] = pcamat(vectors, firstEig, lastEig, interactive, verbose);
Data in row vectors.
Index of the largest eigenvalue to keep. Default is 1.
Index of the smallest eigenvalue to keep. Default is equal to dimension of vectors.
Specify eigenvalues to keep interactively. Note that if you set 'interactive' to 'on' or 'gui' then the values for 'firstEig' and 'lastEig' will be ignored, but they still have to be entered. If the value is 'gui' then the same graphical user interface as in FASTICAG will be used. Default is 'off'.
Default is 'on'.
Calculates the PCA matrices for given data (row) vectors. Returns the eigenvector (E) and diagonal eigenvalue (D) matrices containing the selected subspaces. Dimensionality reduction is controlled with the parameters 'firstEig' and 'lastEig' - but it can also be done interactively by setting parameter 'interactive' to 'on' or 'gui'.
Note The eigenvalues and eigenvectors returned by PCAMAT are not sorted.
For historical reasons this version does not sort the eigenvalues or the eigen vectors in any ways. Therefore neither does the fastica or fasticag. Generally it seams that the components returned from whitening is almost in reversed order. (That means, they usually are, but sometime they are not - depends on the spec-command of scilab.)