This package implements the model selection approach detailed in
McInerney and Burke (2022): “A Statistically-Based Approach to
Feedforward Neural Network Model Selection”. The preprint of this paper
is available on arXiv. More
specifically, the algorithm alternates between selecting the hidden
layer complexity and the inputs with the objective of minimizing the
BIC. The neural network function used is nnet, which is available from
the R package of the same name (Ripley and Venables, 2022).
You can install the development version of selectnn from GitHub with:
# install.packages("devtools")
devtools::install_github("andrew-mcinerney/selectnn")The primary function in this package is selectnn().
library(selectnn)
selectnn(X, y, Q, n_init)
selectnn(y ~ ., data = df, Q, n_init)