NEWS.md

In davidevdt/bayespca: A Package for Estimating PCA with Variational Bayes Inference

bayespca 0.3.0

• previous priors (Jeffrey's, hyperparameters Gamma prior) and SVS functionality have been restored

bayespca 0.2.0

• the package has been heavily revised and modified
• the SVS (Stochastic Variable Selection) functionality has been removed (it might be re-implemented in future versions)
• only fixed and Gamma priors for the precision parameters can be used

bayespca 0.1.0

• removed Type-II-Maximum-Likelihood estimation for SVS inclusion probabilities, as it was causing the algorithm to be too unstable
• updated vignettes
• added variable names to the model output

bayespca 0.0.2

• improved various computational aspects of the VB algorithm
  • more efficient SVD decomposition for large matrices
  • bugs related to the computation of XTX fixed
  • added print-screen iteration info
  • fixed bug related to computation of determinants of too large posterior variance matrices (sigmaWd in expectedvalues.cpp)
• added warning message for positive elbo values in case of unscaled data
• added suppressWarnings argument to hide function warnings
• updatetd line endings of source files to LF
• added BugReports filed in description

davidevdt/bayespca documentation built on Dec. 5, 2020, 3:28 a.m.