qojpca

QOJPCA (Quasi-orthogonal Joint Principal Component Analysis) package allows for improving the orthogonality between linear bases computed through Principal Component Analysis.

Install it from PyPI

The package has been released on PyPip and can be installed through.

pip install qojpca

Requirements

Numpy, Scikit

Usage

The package can be used by importing the "base" module into your python script as:

from qojpca import base
""" 
QOJPCA can be computed using the qojpca static function
l_p (resp. l_q): number of latent variables for P (resp. Q)
l: regularization parameter. Note that it is multiplied by the largest eigenvalue of XX^T
"""
P_vals,Q_vals,P,Q = base.qojpca(X,Y,l_p,l_q,l)

Or by using the command line interface. For example

$ python -m qojpca X.npy Y.npy --l_x 10 --l_y 10 --output_directory output --regularization 100

where: l_p (resp. l_q) are the number of latent variables for linear basis P (resp. Q).