ica_lingam_search: Lingam ICA search

In nicolagnecco/causalXtreme: Causal discovery in heavy-tailed models

Lingam ICA search

Description

Runs ICA-based Lingam given a dataset dat. Returns the causal order estimated from the data using the LiNGAM-ICA algorithm.

Usage

ica_lingam_search(dat, contrast_fun = c("logcosh", "exp"))

Arguments

dat	Numeric matrix. Dataset matrix with n rows (observations) and p columns (variables).
contrast_fun	Character. The functional form of the contrast function used in the Fast-ICA step. It is one of "logcosh" (the default choice) and "exp". For further details see the paper from Hyvarinen, A., https://ieeexplore.ieee.org/abstract/document/761722/.

Details

This function is a wrapper around estLiNGAM from the package pcalg (see https://CRAN.R-project.org/package=pcalg). The function estLiNGAM is slightly modified to allow for different contrast functions in the fast-ICA step of LiNGAM. To modify estLiNGAM, we included in this package several internal functions from pcalg. All the credits go to the authors of the pcalg package:

Markus Kalisch, Alain Hauser , Martin Maechler, Diego Colombo, Doris Entner, Patrik Hoyer, Antti Hyttinen, Jonas Peters, Nicoletta Andri, Emilija Perkovic, Preetam Nandy, Philipp Ruetimann, Daniel Stekhoven, Manuel Schuerch, Marco Eigenmann.

Value

Numeric vector (or NA in case of error). The causal order estimated from the data.

nicolagnecco/causalXtreme documentation built on April 21, 2024, 4:22 a.m.