Struct PredictCohort 

pub struct PredictCohort {
    pub role: PredictCohortRole,
    pub physical_sample_ids: Vec<SampleId>,
    pub origin_sample_ids: Vec<SampleId>,
    pub target_names: Vec<String>,
    pub relation_fingerprint: String,
    pub relations: SampleRelationSet,
    pub data_content_fingerprint: String,
    pub target_content_fingerprint: Option<String>,
    pub cohort_fingerprint: String,
}

Closed, PREDICT-only relation authority introduced by envelope V2.

It is never a supplement to coordinator_relations: CV, OOF, SELECT and training influence continue to consume only the latter. The explicit physical/origin lists prevent a producer from silently changing a cohort by reordering or dropping relation records.

Fields

role: PredictCohortRole

physical_sample_ids: Vec<SampleId>

origin_sample_ids: Vec<SampleId>

target_names: Vec<String>

relation_fingerprint: String

relations: SampleRelationSet

data_content_fingerprint: String

target_content_fingerprint: Option<String>

cohort_fingerprint: String

Implementations

impl PredictCohort

pub fn from_relations( role: PredictCohortRole, relations: SampleRelationSet, target_names: Vec<String>, data_content_fingerprint: String, target_content_fingerprint: Option<String>, ) -> Result<PredictCohort, DagMlError>

Build the closed PREDICT-only authority from its authoritative relation records.

Producers must use this constructor instead of reimplementing the relation, identity and cohort fingerprint algorithms in a host binding. The relation records remain the sole source of physical and origin identities; the derived lists are sorted sets so a producer cannot accidentally make a cohort depend on host iteration order.

pub fn fingerprint(&self) -> Result<String, DagMlError>

pub fn validate(&self) -> Result<(), DagMlError>

pub fn validate_against_cv_fold_set( &self, fold_set: &FoldSet, ) -> Result<(), DagMlError>

External-test rows must never share a physical or origin identity with the CV universe. Inference is intentionally allowed to replay a known sample, but cannot carry targets or selection evidence.

pub fn validate_against_cv_relations( &self, cv_relations: &SampleRelationSet, ) -> Result<(), DagMlError>

Refuse an external-test cohort that intersects the full physical-or-origin identity closure of the CV relation authority.

FoldSet carries only physical sample ids. The coordinator relation set is therefore additionally required to prove that a held-out row did not arrive through a CV origin alias.

Trait Implementations

impl Clone for PredictCohort

fn clone(&self) -> PredictCohort

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PredictCohort

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for PredictCohort

fn deserialize<__D>( __deserializer: __D, ) -> Result<PredictCohort, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for PredictCohort

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

Equality operator ==.

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

Inequality operator !=.

impl Serialize for PredictCohort

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for PredictCohort