rlingam: R implementation of LiNGAM algorithms

Implemented Algorithms

• ICALiNGAM
• DirectLiNGAM

Install

remotes::install_github("gkikuchi/rlingam")

Usage

library(rlingam)

X <- gen_dummy_data(random_state = 10)

# icalingam
mdl <- ICALiNGAM$new()
mdl$fit(X)

# directlingam
mdl <- DirectLiNGAM$new()
mdl$fit(X)

print(mdl$causal_order)
print(mdl$adjacency_matrix)
plot_adjacency_mat(mdl$adjacency_matrix, node_labels = names(X))

Parameters

Parameters for *LiNGAM$new():

• random_state (integer)
  • random seed
• lasso_engine ("glmnet" or "lars")
  • library to use to estimate adjacency matrix. default="glmnet"
• max_iter (integer) only for ICALiNGAM
  • maximum iterations for fastICA. default=1000