huge (0.8)

High-dimensional Undirected Graph Estimation.

The package "huge" provides a general framework for high-dimensional undirected graph estimation. The package integrates data preprocessing (Gaussianization), graph screening, graph estimation, and model selection techniques into a pipeline. The NonparaNormal(NPN) transformation is applied to preprocess the data and helps relax the normality assumption. The Graph SURE Screening (GSS) subroutine preselects the graph neighborhood of each variable. In the graph estimation stage, the structure of either the whole graph or a pre-specified sub-graph is estimated by the Meinshausen & Buhlmann Graph Estimation via Lasso (GEL) strategy on the pre-screened data. In the case d >> n, the computation is memory optimized and is targeted on larger-scale problems (with d>3000). We also provide another efficient method, Graph Approximation via Correlation Thresholding. Three regularization parameter selection methods are included in this package: (1) StARS: Stability Approach for Regularization Selection (2) PIC: Permutation Information Criterion (3) Extended Bayesian Information Criterion (EBIC) based on pseudo-likelihood.

Maintainer: Tuo Zhao
Author(s): Tuo Zhao, Han Liu, Kathryn Roeder, John Lafferty, Larry Wasserman

License: GPL-2

Uses: glmnet, igraph, lattice, MASS, Matrix
Reverse depends: BDgraph, glassomix, ldstatsHD, smart
Reverse suggests: CompareCausalNetworks, edgebundleR, LICORS, pcalg, pulsar, sand, smart, stabs, stm, themetagenomics

Released about 9 years ago.