spherical_pca

Contents

spherical_pca#

QuadratiK.tools.spherical_pca(data: ndarray, scale: bool = False) → dict#

Perform Spherical Principal Component Analysis (PCA). This is a Python implementation of PcaLocantore, the spherical PCA implementation in the R package rrcov.

Parameters#

datanumpy.ndarray

The input data matrix of shape (n_samples, n_features).

scalebool, optional

Whether to scale the data using the Median Absolute Deviation (MAD) before processing. Default is False.

Returns#

resultsdict

A dictionary containing:

  • scoresnumpy.ndarray

    The principal component scores (centered data projected onto the spherical PCA loadings).

  • loadingsnumpy.ndarray

    The principal component loadings (eigenvectors).

  • eigenvaluesnumpy.ndarray

    The estimated eigenvalues (squared MAD of projections).

See Also#

PcaLocantore : Spherical PCA implementation in the R package rrcov. https://search.r-project.org/CRAN/refmans/rrcov/html/PcaLocantore-class.html

sklearn.decomposition.PCA : Standard Principal Component Analysis.

Examples#

import numpy as np
from QuadratiK.tools import spherical_pca
data = np.random.multivariate_normal([0, 0], [[1, 0.5], [0.5, 1]], 100)
results = spherical_pca(data)
print("Keys:", results.keys())
print("Scores shape:", results['scores'].shape)

Keys: dict_keys(['scores', 'loadings', 'eigenvalues'])
Scores shape: (100, 2)
Contents