specmine (version 4.0.0)

multiClassSummary: Multi-class summary metrics

Description

Compute overall and class-averaged performance metrics for multiclass classification models, including ROC AUC and log-loss when class probabilities are available.

Usage

multiClassSummary(data, lev = NULL, model = NULL)

Value

A named numeric vector with overall and class-averaged classification statistics.

Arguments

data

A data frame containing at least the columns `pred` and `obs`, plus one probability column per class when ROC or log-loss are needed.

lev

An optional character vector with the class levels.

model

An optional fitted model object passed by `caret`.

Examples

data <- data.frame(
  pred = factor(c("A", "B", "A", "B"), levels = c("A", "B")),
  obs = factor(c("A", "B", "B", "B"), levels = c("A", "B")),
  A = c(0.8, 0.2, 0.7, 0.3),
  B = c(0.2, 0.8, 0.3, 0.7)
)
multiClassSummary(data)