bozenne/lava.penalty: Penalized Latent Variable Models

Implementation of regularization tools for latent variables models. Available penalty are lasso, ridge, group lasso and nuclear norm regularization (linear model only). A generic regularization path algorithm enable to estimate the knots, i.e. penalisation at which the set of non-zero parameters changes. Information criteria (AIC, BIC) can be used to identify an "optimal" penalization value.

install.packages("remotes")
remotes::install_github("bozenne/lava.penalty")