This documentation is for astroML version 0.2

This page


astroML Mailing List

GitHub Issue Tracker


Scipy 2012 (15 minute talk)

Scipy 2013 (20 minute talk)


If you use the software, please consider citing astroML.

11.7.1. astroML.dimensionality.iterative_pca

astroML.dimensionality.iterative_pca(X, M, n_ev=5, n_iter=15, norm=None, full_output=False)
Parameters :

X: ndarray, shape = (n_samples, n_features) :

input data

M: ndarray, bool, shape = (n_samples, n_features) :

mask for input data. where mask == True, the spectrum is unconstrained

n_ev: int :

number of eigenvectors to use in reconstructing masked regions

n_iter: int :

number of iterations to find eigenvectors

norm: string :

what type of normalization to use on the data. Options are - None : no normalization - ‘L1’ : L1-norm - ‘L2’ : L2-norm

full_output: boolean (optional) :

if False (default) return only the reconstructed data X_recons if True, return the full information (see below)

Returns :

X_recons: ndarray, shape = (n_samples, n_features) :

data with masked regions reconstructed

mu: ndarray, shape = (n_features,) :

mean of data

evecs: ndarray, shape = (min(n_samples, n_features), n_features) :

eigenvectors of the reconstructed data

evals: ndarray, size = min(n_samples, n_features) :

eigenvalues of the reconstructed data

norms: ndarray, size = n_samples :

normalization of each input

coeffs: ndarray, size = (n_samples, n_ev) :

coefficients used to reconstruct X