Struct RunContext 

pub struct RunContext {
    pub run_id: RunId,
    pub root_seed: Option<u64>,
    pub variant_id: Option<VariantId>,
    pub prediction_store: InMemoryPredictionStore,
    pub aggregated_prediction_store: InMemoryAggregatedPredictionStore,
    pub lineage: InMemoryLineageRecorder,
    pub score_collector: Vec<RegressionMetricReport>,
    pub regression_target_records: Vec<RegressionTargetRecord>,
    pub oof_average_blocks: Vec<OofAverageBlock>,
}

Fields

run_id: RunId

root_seed: Option<u64>

variant_id: Option<VariantId>

prediction_store: InMemoryPredictionStore

aggregated_prediction_store: InMemoryAggregatedPredictionStore

lineage: InMemoryLineageRecorder

score_collector: Vec<RegressionMetricReport>

Native per-fold/per-partition score reports collected during the run (when the host emits regression_targets).

regression_target_records: Vec<RegressionTargetRecord>

Per-fold y_true records, kept so cross-fold ensembles (the OOF average) can be scored.

oof_average_blocks: Vec<OofAverageBlock>

The per-sample cross-fold OOF average blocks (+ y_true) collected alongside the scalar OOF average reports — one per scored producer. Surfaced so the host can fill the (validation, avg) row’s per-sample y_pred; populated by collect_cross_fold_validation_scores, empty otherwise.

Implementations

impl RunContext

pub fn new(run_id: RunId, root_seed: Option<u64>) -> Self

pub fn collect_cross_fold_validation_scores( &mut self, partition_mode: FoldPartitionMode, ) -> Result<()>

Score the cross-fold OOF average from the collected per-fold validation predictions + targets and append the reports (one per producer, fold_id = "avg") to the score collector, plus — additively — the per-sample OOF average block + y_true each report was computed from to oof_average_blocks (so the host can fill the (validation, avg) row’s per-sample y_pred). Call after FIT_CV; a no-op when nothing was scored or no producer has more than one fold.

partition_mode is the campaign’s FoldPartitionMode: Partition (KFold) requires a unique per-producer OOF set, while Resampled (ShuffleSplit / repeated CV) permits a sample to be validated in multiple folds (averaged when scored). Pass the plan’s fold_set mode (default Partition when there is no fold set).

pub fn build_score_set( &self, plan_id: impl Into<String>, selection_metric: Option<String>, ) -> Option<ScoreSet>

Build a ScoreSet from the collected reports (or None if scoring was off / produced nothing), e.g. to attach to the ExecutionBundle.

Trait Implementations

impl Clone for RunContext

fn clone(&self) -> RunContext

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RunContext

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

Formats the value using the given formatter.