README.md

In NilsSturma/TestGGM: Testing Gaussian Graphical Models

TestGGM

R package for testing the goodness-of-fit of Gaussian graphical models. Supports two types of models: - Gaussian latent tree models - Two-factor analysis model

Install via devtools::install_github("NilsSturma/TestGGM"). The package implements the test using incomplete U-statistics described in

Sturma, N., Drton, M., and Leung, D. (2022). Testing Many and Possibly Singular Polynomial Constraints. arXiv.

Moreover, it supports similar testing strategies by applying different schemes of forming averages over the kernel estimators.

Most important files: - R/tests.R: algebraic tests for (arbitrary) Gaussian latent tree models - R/LR_EM.R: likelihood-ratio test for (arbitrary) Gaussian latent tree models using the EM algorithm - R/tests-factoranalysis.R algebraic tests for the two-factor analysis model

NilsSturma/TestGGM documentation built on June 30, 2023, 3:09 p.m.