You can install the current version of interpretnn from CRAN with:
install.packages("interpretnn")You can install the development version of interpretnn from GitHub with:
# install.packages("devtools")
devtools::install_github("andrew-mcinerney/interpretnn")The primary function in this package is interpretnn(). It creates a
more statistically-based object of an existing neural network object.
Currently supports neural networks from nnet, neuralnet, keras,
ANN, and torch.
library(interpretnn)
intnn <- interpretnn(object)A useful summary table can be generated using
summary(intnn)and covariate-effect plots can be created using
plot(intnn, conf_int = TRUE)More information about these functions and their arguments can be found in the documentation.
This package is currently in development. If you experience any issues, please get in touch.