Struct BradleyTerryResult 

pub struct BradleyTerryResult {
    pub strengths: HashMap<CandidateId, f64>,
    pub covariance: DMatrix<f64>,
    pub id_to_index: HashMap<CandidateId, usize>,
    pub log_likelihood: f64,
    pub iterations: usize,
    pub converged: bool,
    pub convergence_metric: f64,
}

Result of Bradley-Terry MLE optimization

Scale convention

Both optimization paths (Newton-Raphson and MM) report the point estimate and its uncertainty on the strength scale π = exp(θ):

• strengths holds π_i (strictly positive, mean-centered in log-space so Σ log π_i = 0).
• covariance is Cov(π), i.e. the covariance of the strengths, not of the log-strengths. Newton-Raphson obtains it by the delta method from the sum-to-zero-constrained Fisher information; MM obtains it by bootstrap. Because both are on the same (strength) scale, downstream consumers such as CandidateStats::model_variance and the active-learning acquisition can use them interchangeably.

Fields

strengths: HashMap<CandidateId, f64>

Strength parameters π_i = exp(θ_i) (probability scale, log-strengths sum to zero)

covariance: DMatrix<f64>

Covariance of the strengths Cov(π) (delta-method Fisher⁻¹ or bootstrap)

id_to_index: HashMap<CandidateId, usize>

Mapping from CandidateId to matrix index

log_likelihood: f64

Log-likelihood at solution

iterations: usize

Number of iterations to convergence

converged: bool

Did the algorithm converge?

convergence_metric: f64

Final gradient norm (Newton-Raphson) or max parameter change (MM)

Implementations

impl BradleyTerryResult

pub fn get_estimate(&self, id: CandidateId) -> Option<FitnessEstimate>

Get fitness estimate for a candidate with uncertainty

pub fn all_estimates(&self) -> HashMap<CandidateId, FitnessEstimate>

Get all estimates as a map

pub fn predict_win_probability( &self, a: CandidateId, b: CandidateId, ) -> Option<f64>

Predict probability that candidate a beats candidate b

Trait Implementations

impl Clone for BradleyTerryResult

fn clone(&self) -> BradleyTerryResult

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BradleyTerryResult

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

Formats the value using the given formatter.