Struct LtsFit 

pub struct LtsFit {
    pub coefficients: Array1<f64>,
    pub scale: Scale,
    pub residuals: Array1<f64>,
    pub subset: Vec<usize>,
    pub objective: f64,
    pub coverage: usize,
    pub breakdown_point: f64,
}

A fitted Least Trimmed Squares regression.

LTS is a high-breakdown initializer, not an M-estimator, so (unlike crate::estimator::RegressionFit) it does not implement the ρ-based crate::estimator::RobustEstimator surface. LTS has a well-defined influence function and Gaussian efficiency (it is √n-consistent and asymptotically normal, Rousseeuw & Leroy 1987; Víšek), but these follow from the asymptotics of the trimmed objective, not from ψ(r)/E[ψ'] and 1/V(ψ) (a hard 0/1 trim has no smooth ψ) so they are not reported here. It exposes coefficients, the retained h-subset (so the trimming is auditable), a robust residual scale (MAD) and its coverage-implied breakdown point.

Fields

coefficients: Array1<f64>

Estimated coefficients.

scale: Scale

Robust residual scale (MAD of the residuals).

residuals: Array1<f64>

Residuals y − Xβ̂ for all n observations.

subset: Vec<usize>

Retained subset: the indices of the h smallest-residual observations (ascending).

objective: f64

The LTS objective at the fit: Σ of the h smallest squared residuals.

coverage: usize

Coverage h: the number of observations retained.

breakdown_point: f64

Breakdown point (n − h + 1)/n.

Implementations

impl LtsFit

pub fn coefficients(&self) -> &Array1<f64>

Estimated coefficients.

pub fn scale(&self) -> Scale

Robust residual scale.

pub fn subset(&self) -> &[usize]

The retained h-subset (ascending indices).

pub fn breakdown_point(&self) -> f64

Breakdown point.

Trait Implementations

impl Clone for LtsFit

fn clone(&self) -> LtsFit

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LtsFit

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

Formats the value using the given formatter.