Struct OrdinalFit 

pub struct OrdinalFit { /* private fields */ }

A fitted proportional-odds (ordinal logistic) model for an ordered categorical response with K ≥ 2 levels 0 < 1 < … < K−1.

Uses the cumulative-logit parametrization of R’s MASS::polr:

logit P(yᵢ ≤ k) = αₖ − xᵢᵀβ, k = 0 … K−2,

with strictly increasing thresholds α₀ < α₁ < … < α_{K−2} and a single coefficient vector β shared across all thresholds (the proportional-odds assumption). Positive βⱼ raises the odds of falling in a higher category. The design matrix carries no intercept — the thresholds play that role.

Fit by Newton–Raphson on the full (K−1) + p parameter vector (thresholds then coefficients) with a backtracking line search. With K = 2 this reduces exactly to binary LogisticFit: β equals the logistic slopes and α₀ equals the negated logistic intercept.

Implementations

impl OrdinalFit

pub fn new(x: Array2<f64>, y: Array1<f64>) -> Result<Self>

Fit a proportional-odds model of ordered y on X (default: up to 100 Newton iterations, tolerance 1e-10).

The response holds integer levels 0 … K−1, every level present; K is inferred as max(y) + 1. Do not include an intercept column in X.

Errors

• RegressionError::EmptyInput / RegressionError::ShapeMismatch.
• RegressionError::InvalidResponse for non-integer/negative labels, a missing level, or fewer than two levels.
• RegressionError::RankDeficient if the information matrix is singular.
• RegressionError::NotConverged if Newton’s method fails to converge.

pub fn with_options( x: Array2<f64>, y: Array1<f64>, max_iter: usize, tol: f64, ) -> Result<Self>

Like OrdinalFit::new with an explicit iteration cap and tolerance.

pub fn n_observations(&self) -> usize

Number of observations.

pub fn n_features(&self) -> usize

Number of design columns p.

pub fn n_classes(&self) -> usize

Number of ordered levels K.

pub fn n_parameters(&self) -> usize

Total number of parameters, (K − 1) + p (thresholds plus coefficients).

pub fn iterations(&self) -> usize

Newton iterations taken to converge.

pub fn design_matrix(&self) -> ArrayView2<'_, f64>

The design matrix as fitted.

pub fn response(&self) -> ArrayView1<'_, f64>

The integer level labels.

pub fn thresholds(&self) -> ArrayView1<'_, f64>

Threshold (cutpoint) parameters α₀ < … < α_{K−2}.

pub fn coefficients(&self) -> ArrayView1<'_, f64>

Shared coefficient vector β (proportional-odds effects).

pub fn fitted_probabilities(&self) -> ArrayView2<'_, f64>

Fitted class probabilities, shape n × K.

pub fn covariance(&self) -> ArrayView2<'_, f64>

Covariance of the stacked (K−1)+p parameter vector (thresholds first).

pub fn log_likelihood(&self) -> f64

Maximized log-likelihood.

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

Standard errors of the coefficients β (the last p diagonal entries of the covariance).

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

Standard errors of the thresholds α.

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

Wald z-statistics for the coefficients, βⱼ / seⱼ.

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

Two-sided Wald p-values for the coefficients from the standard normal.

pub fn residual_deviance(&self) -> f64

Residual deviance −2ℓ.

pub fn null_deviance(&self) -> f64

Deviance of the intercept-only (threshold-only, β = 0) model.

pub fn mcfadden_r2(&self) -> f64

McFadden’s pseudo-R², 1 − ℓ/ℓ₀.

pub fn aic(&self) -> f64

Akaike information criterion, −2ℓ + 2·[(K−1)+p].

pub fn bic(&self) -> f64

Bayesian information criterion, −2ℓ + ln(n)·[(K−1)+p].

pub fn predict_proba(&self, x: ArrayView2<'_, f64>) -> Array2<f64>

Predicted class probabilities for a new design matrix x, shape rows × K.

Trait Implementations

impl Clone for OrdinalFit

fn clone(&self) -> OrdinalFit

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for OrdinalFit

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

Formats the value using the given formatter.