Struct ExpectedCalibrationError 

pub struct ExpectedCalibrationError {
    pub num_bins: usize,
}

Expected Calibration Error (ECE) metric.

Measures the difference between predicted probabilities and actual accuracy. ECE divides predictions into bins and computes the average difference between confidence and accuracy across bins, weighted by bin frequency.

Lower ECE indicates better calibration.

Reference: Guo et al. “On Calibration of Modern Neural Networks” (ICML 2017)

Fields

num_bins: usize

Number of bins for calibration

Implementations

impl ExpectedCalibrationError

pub fn new(num_bins: usize) -> Self

Create with custom number of bins.

Trait Implementations

impl Clone for ExpectedCalibrationError

fn clone(&self) -> ExpectedCalibrationError

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ExpectedCalibrationError

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ExpectedCalibrationError

fn default() -> Self

Returns the “default value” for a type.

impl Metric for ExpectedCalibrationError

fn compute( &self, predictions: &ArrayView<'_, f64, Ix2>, targets: &ArrayView<'_, f64, Ix2>, ) -> TrainResult<f64>

Compute metric value.

fn name(&self) -> &str

Get metric name.

fn reset(&mut self)

Reset metric state (for stateful metrics).