bpca.BPCA#

class bpca.BPCA(n_components=None, max_iter=1000, tolerance=0.0001)#

Bayesian principal component analysis (BPCA)

Implements the BPCA method (generative model suggested by Bishop, 1998) as suggested by Oba et al (2003). The implementation follows the reference implementation in R (Stacklies, 2007).

Examples

from bpca import BPCA
from sklearn.datasets import load_iris

iris_dataset = load_iris()
X = iris_dataset["data"] # (n_obs, n_var)
bpca = BPCA(n_components=None)

usage = bpca.fit_transform(X) # (n_components, n_latent)
weights = bpca.components_  # (n_latent, n_var)

Citation#

Bishop, C. Bayesian PCA. in Advances in Neural Information Processing Systems vol. 11 (MIT Press, 1998).
Oba, S. et al. A Bayesian missing value estimation method for gene expression profile data. Bioinformatics 19, 2088 - 2096 (2003).
Stacklies, W., Redestig, H., Scholz, M., Walther, D. & Selbig, J. pcaMethods—a bioconductor package providing PCA methods for incomplete data. Bioinformatics 23, 1164 - 1167 (2007).

Attributes table#

`alpha`	Estimated regularization strength $\alpha^2$ by EM algorithm.
`components_`	Principal axes in feature space (`n_components_`, `n_features`).
`explained_variance_ratio_`	Fraction of variance explained by each of the selected components (`n_components_`,).
`mu`	Estimated feature-wise mean $\mu$ by EM algorithm.
`n_components_`	Number of components (`n_components_`,).
`n_iter`	Number of iterations
`tau`	Estimated variance $\sigma^2$ by EM algorithm.

Methods table#

`fit`(X)	Fit parameters.
`fit_transform`(X)	Fit model and return transformed training data.
`transform`(X)	Transform new data using fitted model.

Attributes#

BPCA.alpha#: Estimated regularization strength $\alpha^2$ by EM algorithm.

BPCA.components_#: Principal axes in feature space (n_components_, n_features).

BPCA.explained_variance_ratio_#: Fraction of variance explained by each of the selected components (n_components_,).

BPCA.mu#: Estimated feature-wise mean $\mu$ by EM algorithm.

BPCA.n_components_#: Number of components (n_components_,).

BPCA.n_iter#: Number of iterations

BPCA.tau#: Estimated variance $\sigma^2$ by EM algorithm.

Methods#

BPCA.fit(X)#

Fit parameters.

Parameters:: X (ndarray) – Data matrix (n_obs, n_var)
Return type:: BPCA
Returns:: BPCA Fitted model (self)

BPCA.fit_transform(X)#

Fit model and return transformed training data.

Parameters:: X (ndarray) – Data matrix (n_obs, n_var)
Return type:: ndarray
Returns:: np.ndarray Transformed data (n_obs, n_components)

BPCA.transform(X)#

Transform new data using fitted model.

Parameters:: X (ndarray) – Data matrix (n_obs, n_var)
Return type:: ndarray
Returns:: np.ndarray Transformed data (n_obs, n_components)

bpca.BPCA

Contents