Enum RegressionLoss 

#[non_exhaustive]
pub enum RegressionLoss {
    SquaredError,
    AbsoluteError,
    Huber {
        alpha: f64,
    },
    Quantile {
        alpha: f64,
    },
}

Loss function for gradient boosting regression.

Controls how pseudo-residuals are computed and how leaf values are determined. Different losses provide different robustness properties.

• SquaredError (default): standard MSE loss, optimal for Gaussian noise.
• AbsoluteError: L1 loss (MAE), more robust to outliers.
• Huber { alpha }: hybrid of squared and absolute error; alpha is the quantile at which the transition occurs (default 0.9).
• Quantile { alpha }: predicts the alpha-quantile of the conditional distribution (default 0.5 = median).

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

SquaredError

Least-squares loss (MSE). Default.

AbsoluteError

Least-absolute-deviation loss (MAE).

Huber

Huber loss — squared error for small residuals, absolute for large. alpha is the quantile threshold (typically 0.9).

Fields

alpha: f64

Quantile threshold for switching between squared and absolute.

Quantile

Quantile loss — predicts the alpha-quantile. alpha = 0.5 gives the median.

Fields

alpha: f64

Target quantile in (0, 1).

Trait Implementations

impl Clone for RegressionLoss

fn clone(&self) -> RegressionLoss

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RegressionLoss

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

Formats the value using the given formatter.

impl Default for RegressionLoss

fn default() -> RegressionLoss

Returns the “default value” for a type.