Struct ExpectationMaximization 

pub struct ExpectationMaximization {
    pub n_components: usize,
    pub components: Vec<GmmComponent>,
    pub ll_history: Vec<f64>,
    pub tol: f64,
}

Expectation-Maximization for Gaussian Mixture Models.

Supports k-means initialization, BIC criterion for model selection, and full EM convergence.

Fields

n_components: usize

Number of components.

components: Vec<GmmComponent>

Mixture components.

ll_history: Vec<f64>

Log-likelihood history.

tol: f64

Convergence tolerance.

Implementations

impl ExpectationMaximization

pub fn new(n_components: usize) -> Self

Creates a new ExpectationMaximization with k-means seeding.

pub fn with_tol(self, tol: f64) -> Self

Sets convergence tolerance.

pub fn kmeans_init(&mut self, data: &[f64])

Initializes component means via k-means (one pass, sorted data).

pub fn log_likelihood(&self, data: &[f64]) -> f64

Computes log-likelihood of data under current model.

pub fn bic(&self, data: &[f64]) -> f64

Computes the Bayesian Information Criterion.

BIC = k ln(n) - 2 ln(L̂)

pub fn fit(&mut self, data: &[f64], max_iter: usize) -> f64

Runs the EM algorithm.

Returns the final log-likelihood.

pub fn predict(&self, x: f64) -> usize

Predicts cluster assignment (most likely component) for a data point.

pub fn normalized_weights(&self) -> Vec<f64>

Returns component weights normalized to sum to 1.

Trait Implementations

impl Clone for ExpectationMaximization

fn clone(&self) -> ExpectationMaximization

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ExpectationMaximization

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

Formats the value using the given formatter.