trainingReport: Training history evaluation of a DNN model

In SEMdeep: Structural Equation Modeling with Deep Neural Network and Machine Learning Algorithms

Training history evaluation of a DNN model

Description

Display a (r,c) panel history plot from SEMdnn() output, with x = number of epochs, y = training loss for each MLP model and bootstrap sample, if nboot > 0.

Usage

trainingReport(object, size = NULL, r = 2, c = 2, ...)

Arguments

object	A model fitting object from SEMdnn() function.
size	number of the multiple plots (default, size = NULL: all training MLP for each bootstrap sample are visualized).
r	number of rows of the plot layout (default, r = 2).
c	number of columns of the plot layout (default, c = 2).
...	Currently ignored.

Details

The training history plot can provide an indication about the training of the model, such as: (i) its speed of convergence over epochs (slope), (ii) whether the model may have already converged (plateau of the line), (iii) whether the mode may be over-learning the training data (inflection for validation line), and more.

Value

Invisibly returns a data.frame of the training loss history.

Author(s)

Mario Grassi [e-mail]

Examples

if (torch::torch_is_installed()){

# Load Sachs data (pkc)
ig<- sachs$graph
data<- sachs$pkc
data<- transformData(data)$data
group<- sachs$group

#...with train-test (0.5-0.5) samples
set.seed(123)
train<- sample(1:nrow(data), 0.5*nrow(data))

dnn <- SEMdnn(ig, data[train, ], algo = "layerwise",
              hidden = 10, link = "relu", loss = "mse",
              validation = 0.2, nboot = 0, epochs = 32)

tr <- trainingReport(dnn); tr
}

SEMdeep documentation built on Dec. 12, 2025, 1:07 a.m.