Struct EvidenceEntry 

pub struct EvidenceEntry {
    pub decision_id: u64,
    pub timestamp_ns: u64,
    pub domain: DecisionDomain,
    pub log_posterior: f64,
    pub top_evidence: [Option<EvidenceTerm>; 3],
    pub action: &'static str,
    pub loss_avoided: f64,
    pub confidence_interval: (f64, f64),
}

Unified evidence record for any Bayesian decision point.

Fixed-size: the top-3 evidence array avoids heap allocation.

Fields

decision_id: u64

Monotonic decision counter (unique within a session).

timestamp_ns: u64

Monotonic timestamp (nanoseconds from program start).

domain: DecisionDomain

Which decision domain this belongs to.

log_posterior: f64

Log-posterior odds of the chosen action being optimal.

top_evidence: [Option<EvidenceTerm>; 3]

Top-3 evidence terms ranked by |log(BF)|, pre-allocated.

action: &'static str

Action taken (e.g., “dirty_rows”, “coalesce”, “degrade_1”).

loss_avoided: f64

Expected loss avoided: E[loss(next_best)] - E[loss(chosen)]. Non-negative when the chosen action is optimal.

confidence_interval: (f64, f64)

Confidence interval (lower, upper) on the posterior probability.

Implementations

impl EvidenceEntry

pub fn posterior_probability(&self) -> f64

Posterior probability derived from log-odds.

pub fn evidence_count(&self) -> usize

Number of evidence terms present.

pub fn combined_log_bf(&self) -> f64

Combined log Bayes factor (sum of individual log-BFs).

pub fn to_jsonl(&self) -> String

Format as a JSONL line (no trailing newline).

Trait Implementations

impl Clone for EvidenceEntry

fn clone(&self) -> EvidenceEntry

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EvidenceEntry

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

Formats the value using the given formatter.