Struct MixedModel 

pub struct MixedModel { /* private fields */ }

A fitted general linear mixed model with one or more random-effect terms — random slopes and/or crossed & nested grouping factors — estimated by REML (default) or ML.

The model is y = Xβ + Zb + ε, b ~ N(0, G), ε ~ N(0, σ²_e I), where Z and G are assembled from the supplied RandomEffect terms (G is block-diagonal, repeating each term’s k × k covariance across its groups). Estimation profiles β (by GLS) and σ²_e out analytically and optimizes the remaining relative covariance parameters with a Nelder–Mead search, using a dense Cholesky solve of the n × n marginal covariance at each step.

For the single random-intercept case prefer the closed-form LinearMixedModel; this type handles everything beyond it, and reduces to it exactly for one intercept term.

Scale

The dense solve is O(n³) — appropriate for the grouped datasets these diagnostics target, not for very large n.

Implementations

impl MixedModel

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

Fit by REML. X holds the fixed effects (intercept included); terms are the random-effect terms.

Errors

• RegressionError::EmptyInput / RegressionError::ShapeMismatch.
• RegressionError::NoResidualDegreesOfFreedom if n ≤ p.
• RegressionError::InvalidResponse if no terms are given.
• RegressionError::RankDeficient if the GLS system is singular.

pub fn with_method( x: Array2<f64>, y: Array1<f64>, terms: Vec<RandomEffect>, method: Method, ) -> Result<Self>

Like MixedModel::new with an explicit Method.

pub fn n_observations(&self) -> usize

Number of observations.

pub fn n_parameters(&self) -> usize

Number of fixed-effect coefficients.

pub fn n_random_effects(&self) -> usize

Total number of random-effect coefficients across all terms.

pub fn n_terms(&self) -> usize

Number of random-effect terms.

pub fn method(&self) -> Method

Estimation method used.

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

Fixed-effect coefficients β̂.

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

Fixed-effect covariance σ̂²_e (XᵀV⁻¹X)⁻¹.

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

Fixed-effect standard errors.

pub fn residual_variance(&self) -> f64

Residual (within-group) variance σ̂²_e.

pub fn term_covariance(&self, t: usize) -> ArrayView2<'_, f64>

Estimated covariance matrix Σ̂ (k × k) of random-effect term t; the diagonal holds the intercept/slope variances, the off-diagonal their covariance.

pub fn random_effects(&self, t: usize) -> ArrayView2<'_, f64>

BLUPs for random-effect term t, shape n_groups × k (row = group, column = the term’s component).

pub fn log_likelihood(&self) -> f64

The profile log-likelihood (REML or ML) at the estimate.

pub fn aic(&self) -> f64

AIC. Under ML the parameter count is p plus the number of covariance parameters plus one for σ²_e; under REML only the covariance parameters and σ²_e are counted.

Trait Implementations

impl Clone for MixedModel

fn clone(&self) -> MixedModel

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MixedModel

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

Formats the value using the given formatter.