](https://github.com/citoverse/cito/actions)
](https://github.com/citoverse/cito/actions)
‘cito’ aims at helping you build and train Neural Networks with the standard R syntax. It allows the whole model creation process and training to be done with one line of code. Furthermore, all generic R methods such as print or plot can be used on the created object. It is based on the ‘torch’ machine learning framework which is available for R. Since it is native to R, no Python installation or any further API is needed for this package.
Before installing cito make sure torch is installed.
install.packages('torch')
The development version from GitHub is available with:
# install.packages('devtools')
devtools::install_github('citoverse/cito')Once installed, the main function dnn() can be used. See the example below. A more in depth example with more explanation can be found in the vignette.
library(cito)
validation_set <- sample(c(1:nrow(datasets::iris)),25)
# Build and train Network
nn.fit <- dnn(Sepal.Length~., data = datasets::iris[-validation_set,])
# Analyze training
analyze_training(nn.fit)
# Print sturcture of Neural Network
print(nn.fit)
# Plot Structure of Neural Network
plot(nn.fit)
# continue training for another 32 epochs
nn.fit< - continue_training(nn.fit)
# Use model on validation set
predictions <- predict(nn.fit, iris[validation_set,])
# Scatterplot
plot(iris[validation_set,]$Sepal.Length,predictions)
# MAE
mean(abs(predictions-iris[validation_set,]$Sepal.Length))