Conformal Prediction

Conformal-prediction reporting stability helpers.

class updatesupport.conformal.ConformalTargetSpec(column, label, role, target_description, ambiguity_limit=None)[source]

Bases: object

One conformal-derived aggregate target to audit.

Parameters:

  • column (str)

  • label (str)

  • role (str)

  • target_description (str)

  • ambiguity_limit (float | None)

column: str
label: str
role: str
target_description: str
ambiguity_limit: float | None = None
classmethod from_value(value, *, ambiguity_limit=None)[source]

Build a conformal target spec from a string, mapping, or spec.

Parameters:
Return type:

ConformalTargetSpec

as_dict()[source]
Return type:

dict[str, Any]

class updatesupport.conformal.ConformalTargetAudit(spec, claim, design)[source]

Bases: object

Audit output for one conformal-derived target column.

Parameters:
spec: ConformalTargetSpec
claim: ClaimSpec
design: PublicReportDesign
property audit: ClaimAudit
property status: str
property design_status: str
property observed_value: float
property lower: float
property upper: float
property ambiguity: float
property public_adequate: bool
property recommended_public: tuple[str, ...] | None
property recommended_label: str | None
as_dict()[source]
Return type:

dict[str, Any]

class updatesupport.conformal.ConformalReportingStabilityReport(target_audits, public, hidden, weight=None, source='conformal', source_rows=None, metadata=None, title='Conformal Reporting Stability')[source]

Bases: ReportArtifactMixin

Multi-target stability report for conformal-prediction outputs.

Parameters:
target_audits: tuple[ConformalTargetAudit, ...]
public: tuple[str, ...]
hidden: tuple[str, ...]
weight: str | None = None
source: str = 'conformal'
source_rows: int | None = None
metadata: Mapping[str, Any] | None = None
title: str = 'Conformal Reporting Stability'
property target_count: int
property pass_count: int
property fail_count: int
property inconclusive_count: int
property needs_refinement_count: int
property status: str
as_dict()[source]
Return type:

dict[str, Any]

to_tables()[source]
Return type:

dict[str, tuple[dict[str, Any], …]]

to_markdown()[source]
Return type:

str

updatesupport.conformal.conformal_reporting_stability(data, *, public, hidden, weight=None, targets=None, candidate_refinements=(), ambiguity_limit=None, ambiguity_limits=None, q=None, q_presets=('saturated',), min_cell_weight=1.0, top=10, include_attribution=False, title='Conformal Reporting Stability', **claim_kwargs)[source]

Audit useful conformal-derived aggregate targets in one report.

data may be a ConformalAdapterResult or ordinary row-like data. When adapter metadata is available, target columns are discovered automatically. Otherwise pass targets explicitly.

Parameters:
Return type:

ConformalReportingStabilityReport