MXM: Feature Selection (Including Multiple Solutions) and Bayesian Networks

Many feature selection methods for a wide range of response variables, including minimal, statistically-equivalent and equally-predictive feature subsets. Bayesian network algorithms and related functions are also included. The package name 'MXM' stands for "Mens eX Machina", meaning "Mind from the Machine" in Latin. Reference: Feature Selection with the R Package MXM: Discovering Statistically Equivalent Feature Subsets, Lagani, V. and Athineou, G. and Farcomeni, A. and Tsagris, M. and Tsamardinos, I. (2017). Journal of Statistical Software, 80(7). <doi:10.18637/jss.v080.i07>.

Version: 1.4.2
Depends: R (≥ 3.5)
Imports: methods, stats, utils, survival, MASS, graphics, ordinal, nnet, quantreg, lme4, foreach, doParallel, parallel, relations, Rfast, visNetwork, energy, geepack, knitr, dplyr, bigmemory, coxme
Suggests: R.rsp
Published: 2019-01-15
Author: Michail Tsagris [aut, cre], Ioannis Tsamardinos [aut, cph], Vincenzo Lagani [aut, cph], Giorgos Athineou [aut], Giorgos Borboudakis [ctb], Anna Roumpelaki [ctb]
Maintainer: Michail Tsagris <mtsagris at>
License: GPL-2
NeedsCompilation: no
Citation: MXM citation info
In views: MachineLearning, gR
CRAN checks: MXM results


Reference manual: MXM.pdf
Vignettes: Guide on performing feature selection with the R package MXM
Discovering Statistically-Equivalent Feature Subsets with MXM
A very brief guide to using MXM
Tutorial: Feature selection with the FBED algorithm
Tutorial: Feature selection with the MMPC algorithm
Tutorial: Bayesian Network Construction
Tutorial: Feature selection with the SES algorithm
Package source: MXM_1.4.2.tar.gz
Windows binaries: r-devel:, r-release:, r-oldrel:
OS X binaries: r-release: MXM_1.4.2.tgz, r-oldrel: MXM_1.4.2.tgz
Old sources: MXM archive


Please use the canonical form to link to this page.