Enum DispersionFamilyKind 

pub enum DispersionFamilyKind {
    NegativeBinomial,
    Gamma,
    Beta,
    Tweedie {
        p: f64,
    },
}

The genuine-dispersion mean family whose precision (overdispersion) channel can carry a second noise_formula linear predictor (issue #913).

Variants

NegativeBinomial

NB2: Var = μ + μ²/θ; the precision channel models log θ.

Gamma

Gamma with Var = μ²/ν; the precision channel models log ν (shape).

Beta

Beta(μφ, (1−μ)φ) with a logit mean link; the precision channel models log φ.

Tweedie

Tweedie compound Poisson–Gamma with Var = φ μ^p, fixed power p; the precision channel models log(1/φ). The per-row density uses the saddlepoint (Nelder–Pregibon) approximation for y > 0 and the exact point mass at y = 0; this is the standard tractable Tweedie ML surface (an exact-series φ-derivative is the remaining hard sub-item of #913).

Fields

p: f64

Implementations

impl DispersionFamilyKind

pub const fn family_tag(self) -> &'static str

pub fn base_link(self) -> InverseLink

The mean inverse link this dispersion family fits on: log for NegativeBinomial / Gamma / Tweedie, logit for Beta. Single source of truth shared by the CLI and FFI save paths so the persisted base_link never diverges from the fitted channel.

pub fn likelihood_spec(self) -> LikelihoodSpec

The family’s canonical [LikelihoodSpec] (mean response × mean link). The overdispersion parameter is estimated by the log-precision channel, so the response-family placeholder parameters (phi, theta) mirror the resolve_family defaults and are not consumed as fixed values at predict time. This is the single source of truth for the persisted location-scale likelihood so the CLI and FFI save paths cannot diverge.

Trait Implementations

impl Clone for DispersionFamilyKind

fn clone(&self) -> DispersionFamilyKind

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for DispersionFamilyKind

impl Debug for DispersionFamilyKind

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

Formats the value using the given formatter.

impl PartialEq for DispersionFamilyKind

fn eq(&self, other: &DispersionFamilyKind) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for DispersionFamilyKind