localModel: Local Explanations of Machine Learning Models for Tabular Data.
localModel is a successor to the
live package. It implements a variant of LIME method for explaining single predictions of black box machine learning models for tabular data.
Interpretable features are created based on Ceteris Paribus plots.
Details of the methodology are described in the vignette.
localModel is currently undergoing rapid changes, including bug fixes, for a stable solution please see the
To get started, install the package from CRAN:
The development version can be installe from GitHub by using the code below.
Please do not use the
devtools package, since it is affected a bug which makes
localModel installation impossible.
This issue was solved in the Github version of
To get help, see examples and details of the methodology, please refer to package website and vignettes.
Work on this package is financially supported by the NCN Opus grant 2017/27/B/ST6/01307.