bayespca: Regularized Principal Component Analysis via Variational Bayes inference

An R package for regularized Principal Component Analysis via Variational Bayes methods.

Author

Davide Vidotto d.vidotto@uvt.nl

Description

bayespca performs Bayesian estimation of weight vectors in PCA.
To achieve regularization, the method allows specifying fixed precisions
in the prior distributions of the weights; alternatively, it is possible
to implement Gamma priors on such parameters. The method allows
for variable selection through Automatic Relevance Determination.
Check the vignettes and package documentation for further details.

Functions

• vbpca for model estimation
• vbpca_control for settings of control parameters
• is.vbpca for testing the class
• plotheatmap for plotting the precision and weights matrices;
• plothpdi for plotting high probability density intervals

Install

devtools::install_github(“davidevdt/bayespca”)

Version

0.3.0

Depends

R (>= 3.3.3)

License

GPL-2