ICALiNGAM: ICALiNGAM class

In gkikuchi/rlingam: R Implementation of LiNGAM algorithms

Description

R implementation of ICA based LiNGAM algorithm See reference for details of the algorithm

Super class

rlingam::LiNGAM -> ICALiNGAM

Public fields

max_iter

(integer) maximum iterations for fastICA

Methods

Public methods

• ICALiNGAM$new()

• ICALiNGAM$fit()

• ICALiNGAM$estimate_causal_order()

• ICALiNGAM$search_causal_order()

• ICALiNGAM$clone()

Inherited methods

• rlingam::LiNGAM$estimate_adjacency_matrix()

• rlingam::LiNGAM$predict_adaptive_lasso()

---

Method new()

create ICALiNGAM object

Usage

ICALiNGAM$new(random_state = NULL, lasso_engine = "glmnet", max_iter = 1000)

Arguments

random_state

(integer) Random seed

lasso_engine

(character) "lars" or "glmnet"

max_iter

(integer) maximum iterations of fastICA

---

Method fit()

fit DirectLiNGAM

Usage

ICALiNGAM$fit(X)

Arguments

X

(numeric matrix or data.frame) data matrix to fit Estimate causal order based on estimated adjacency matrix

---

Method estimate_causal_order()

if B is not DAG, set small elements to zero until DAG is obtained

Usage

ICALiNGAM$estimate_causal_order(B)

Arguments

B

(matrix) estimated adjacency matrix

Returns

integer vector of length(ncol(B)) get causal order of estimated adjacency matrix

---

Method search_causal_order()

if given matrix is not DAG, return NULL

Usage

ICALiNGAM$search_causal_order(B)

Arguments

B

(matrix) estimated adjacency matrix

Returns

vector of length(ncol(B)) or NULL

---

Method clone()

The objects of this class are cloneable with this method.

Usage

ICALiNGAM$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

References

S. Shimizu, P. O. Hoyer, A. Hyvärinen, and A. J. Kerminen. A linear non-gaussian acyclic model for causal discovery. Journal of Machine Learning Research, 7:2003-2030, 2006.

gkikuchi/rlingam documentation built on Jan. 7, 2022, 11:10 p.m.