/bayespcaSim

Simulation Study for the Bayes PCA method

Primary LanguageR

bayespcaSim

Simulation Study for the Bayes PCA method

The method (Bayes PCA) is compared with the sparse PCA specification of the model. Simulations are run with the R (>= 3.3.0) programming language. Bayesian PCA is implemented via the bayespca package by D. Vidotto (https://github.com/davidevdt/bayespca), while Sparse PCA is implemented with the elasticnet package by H. Zou (https://github.com/cran/elasticnet).

Run the simulations

To run the simulations:

  1. Install the required R packages:
    • devtools::install_github("cran/elasticnet")
    • devtools::install_github("davidevdt/bayespca")
  2. Launch main.R ; in this file, simulation parameters and plotting functions can be specified
    • modify the simulation parameters by changing the values that appear before runSim()
    • select the type of results you want to visualize: plotN = 1 for Tucker congruence, plotN = 2 for proportion of correct zeros/nonzeros, plotN = 3 for the reconstruction errors