Nothing Special   »   [go: up one dir, main page]

tabnet: Fit 'TabNet' Models for Classification and Regression

Implements the 'TabNet' model by Sercan O. Arik et al. (2019) <doi:10.48550/arXiv.1908.07442> with 'Coherent Hierarchical Multi-label Classification Networks' by Giunchiglia et al. <doi:10.48550/arXiv.2010.10151> and provides a consistent interface for fitting and creating predictions. It's also fully compatible with the 'tidymodels' ecosystem.

Version: 0.6.0
Depends: R (≥ 3.6)
Imports: coro, data.tree, dials, dplyr, ggplot2, hardhat (≥ 1.3.0), magrittr, Matrix, methods, parsnip, progress, purrr, rlang, stats, stringr, tibble, tidyr, torch (≥ 0.4.0), tune, utils, vctrs, withr, zeallot
Suggests: knitr, modeldata, patchwork, recipes, rmarkdown, rsample, spelling, testthat (≥ 3.0.0), tidymodels, tidyverse, vip, visdat, workflows, yardstick
Published: 2024-06-15
DOI: 10.32614/CRAN.package.tabnet
Author: Daniel Falbel [aut], RStudio [cph], Christophe Regouby [cre, ctb], Egill Fridgeirsson [ctb], Philipp Haarmeyer [ctb], Sven Verweij ORCID iD [ctb]
Maintainer: Christophe Regouby <christophe.regouby at free.fr>
BugReports: https://github.com/mlverse/tabnet/issues
License: MIT + file LICENSE
URL: https://mlverse.github.io/tabnet/, https://github.com/mlverse/tabnet
NeedsCompilation: no
Language: en-US
Materials: README NEWS
CRAN checks: tabnet results

Documentation:

Reference manual: tabnet.pdf
Vignettes: Hierarchical Classification
Training a Tabnet model from missing-values dataset
Interpretation tools
Self-supervised training and fine-tuning
Fitting tabnet with tidymodels

Downloads:

Package source: tabnet_0.6.0.tar.gz
Windows binaries: r-devel: tabnet_0.6.0.zip, r-release: tabnet_0.6.0.zip, r-oldrel: tabnet_0.6.0.zip
macOS binaries: r-release (arm64): tabnet_0.6.0.tgz, r-oldrel (arm64): tabnet_0.6.0.tgz, r-release (x86_64): tabnet_0.6.0.tgz, r-oldrel (x86_64): tabnet_0.6.0.tgz
Old sources: tabnet archive

Linking:

Please use the canonical form https://CRAN.R-project.org/package=tabnet to link to this page.