plot_roc: Plot Classification ROC Curves

In rtemis: Schema-Driven Machine Learning and Visualization

Plot Classification ROC Curves

Description

S7 generic for ordinary and resampled rtemis classification objects using ECharts. All available training, validation, and test probability samples are selected by default. Resampled results use training and test. An explicit unavailable sample is an error; absence of training probabilities does not hide test data.

Usage

plot_roc(x, ...)

Arguments

x	rtemis classification object: Ordinary or resampled result.
...	Method arguments, including what (Character: "all" or unique sample names), variant (Character: "aggregate" or "per_resample"), positive (Optional Character: binary positive class), and labelify (Logical: capitalize sample labels), followed by additional named arguments to rtemis.draw::draw_roc().

Details

variant = "aggregate" pools aligned prediction rows, retaining repeated observations. "per_resample" draws individual curves with mean and sample SD of fold AUCs. Missing probability folds remain undefined curves and are disclosed; they are not zero AUCs. All curves use the same named classes and binary positive class, defaulting to the first fold's second outcome level.

Value

An ECharts htmlwidget.

Examples

model <- train(iris, hyperparameters = setup_CART(), verbosity = 0L)
plot_roc(model)

rtemis documentation built on Oct. 20, 2026, 1:06 a.m.