vivo: Local Variable Importance via Oscillations of Ceteris Paribus Profiles

Provides an easy to calculate variable importance measure based on Ceteris Paribus plot and is calculated in eight variants. We obtain eight variants measure through the possible combinations of three parameters such as absolute_deviation, point and density.

Version: 0.1.1
Depends: R (≥ 3.0)
Imports: ggplot2, dplyr, ingredients, DALEX
Suggests: knitr, rmarkdown, mlbench, randomForest, gridExtra, grid, lattice, testthat
Published: 2019-09-03
Author: Anna Kozak [aut, cre], Przemyslaw Biecek [aut, ths]
Maintainer: Anna Kozak <anna1993kozak at gmail.com>
BugReports: https://github.com/ModelOriented/vivo/issues
License: GPL-2
URL: https://github.com/ModelOriented/vivo
NeedsCompilation: no
Materials: README NEWS
CRAN checks: vivo results

Downloads:

Reference manual: vivo.pdf
Vignettes: Example with random forest model
Package source: vivo_0.1.1.tar.gz
Windows binaries: r-devel: vivo_0.1.1.zip, r-release: vivo_0.1.1.zip, r-oldrel: vivo_0.1.1.zip
OS X binaries: r-release: vivo_0.1.1.tgz, r-oldrel: vivo_0.1.1.tgz
Old sources: vivo archive

Linking:

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