Package: mig 1.0

Leo Belzile

mig: Multivariate Inverse Gaussian Distribution

Provides utilities for estimation for the multivariate inverse Gaussian distribution of Minami (2003) <doi:10.1081/STA-120025379>, including random vector generation and explicit estimators of the location vector and scale matrix. The package implements kernel density estimators discussed in Belzile, Desgagnes, Genest and Ouimet (2024) <doi:10.48550/arXiv.2209.04757> for smoothing multivariate data on half-spaces.

Authors:Frederic Ouimet [aut], Leo Belzile [aut, cre]

mig_1.0.tar.gz
mig_1.0.tar.gz(r-4.5-noble)mig_1.0.tar.gz(r-4.4-noble)
mig_1.0.tgz(r-4.4-emscripten)mig_1.0.tgz(r-4.3-emscripten)
mig.pdf |mig.html✨
mig/json (API)

# Install 'mig' in R:
install.packages('mig', repos = 'https://cloud.r-project.org')

Bug tracker:https://github.com/lbelzile/mig/issues0 issues

Uses libs:
  • openblas– Optimized BLAS
  • c++– GNU Standard C++ Library v3
  • openmp– GCC OpenMP (GOMP) support library
Datasets:

On CRAN:

Conda:

openblascppopenmp

2.70 score 165 downloads 19 exports 9 dependencies

Last updated 8 months agofrom:c63d5ab735. Checks:3 OK. Indexed: no.

TargetResultLatest binary
Doc / VignettesOKMar 12 2025
R-4.5-linux-x86_64OKMar 12 2025
R-4.4-linux-x86_64OKMar 12 2025

Exports:.lsum.mig_mle.mig_momdmigdmig_laplaciandtelliptfit_migmig_kdensmig_kdens_bandwidthmig_lcvmig_loglik_gradmig_loglik_hessianmig_loglik_laplacianmig_rlcvnormalrule_bandwidthpmigrmigrtellipttellipt_kdens

Dependencies:alabamanleqslvnumDerivqrngRcppRcppArmadillospacefillrstatmodTruncatedNormal

mig package

Rendered frommig_vignette.Rmdusingknitr::rmarkdownon Mar 12 2025.

Last update: 2024-07-15
Started: 2024-07-15

Citation

To cite package ‘mig’ in publications use:

Ouimet F, Belzile L (2024). mig: Multivariate Inverse Gaussian Distribution. R package version 1.0, https://CRAN.R-project.org/package=mig.

Corresponding BibTeX entry: 

  @Manual{,
    title = {mig: Multivariate Inverse Gaussian Distribution},
    author = {Frederic Ouimet and Leo Belzile},
    year = {2024},
    note = {R package version 1.0},
    url = {https://CRAN.R-project.org/package=mig},
  }

Readme and manuals

Multivariate inverse Gaussian

This R package consists of utilities for multivariate inverse Gaussian (MIG) models with mean $\boldsymbol{\xi}$ and scale matrix $\boldsymbol{\Omega}$ defined over the halfspace ${\boldsymbol{x} \in \mathbb{R}^d: \boldsymbol{\beta}^\top\boldsymbol{x} > 0}$, including density evaluation and random number generation and kernel smoothing.

Distributions
  • mig for the MIG distribution(rmig for random number generation and dmig for density)
  • tellipt (rtellipt for random vector generation and dtellipt the density) for truncated Student-$t$ or Gaussian distribution over the half space ${\boldsymbol{x}: \boldsymbol{\beta}^\top\boldsymbol{x}>\delta}$ for $\delta \geq 0$.
  • fit_mig to estimate the parameters of the MIG distribution via maximum likelihood (mle) or the method of moments (mom).
Kernel density estimation
  • mig_kdens_bandwidth to estimate the bandwidth matrix minimizing the asymptotic mean integrated squared error (AMISE) or the leave-one-out likelihood cross validation, minimizing the Kullback--Leibler divergence. The amise estimators are estimated by drawing from a mig or truncated Gaussian vector via Monte Carlo
  • normalrule_bandwidth for the normal rule of Scott for the Gaussian kernel
  • mig_kdens for the kernel density estimator
  • tellipt_kdens for the truncated Gaussian kernel density estimator