API reference

Input panel

The one input format and its validation.

validate_estimates	Check an input panel and return it in a canonical order.
validate_readings	Check a frame of reading moments against the panel it will be read from.
PanelError	An input panel does not meet the format every method assumes.
pivot_sources	Turn one anchored row per source into one row per subject.
subject_index	Return the columns that identify a subject in this frame.
available_sources	Return the source names present in a panel, in alphabetical order.

Point-in-time reading

latest_before	Return each source’s latest estimate per subject as of one moment.
attach_latest_before	Attach to each row the latest observation known strictly before its own moment.
first_known	Keep the first-known observation per key, dropping every later revision.
age_at	Add a source’s staleness, as a value, beside each source’s known_at.

History statistics

consensus_revision	Compute how far the consensus has moved over a lookback window.
consensus_bias	Estimate each entity’s persistent consensus error from its own history.
growth_prior	Compute the prior on an entity’s own growth from its realised history.
growth_error_variance	Compute one source’s growth track record ahead of an entity’s history.

Anchored family

Publication, the estimate, its coefficient sets and eligibility.

publish_anchored	Publish tidy anchored estimates under the 30-day revision policy.
prepare_anchored_subjects	Anchor selected sources into stable forecast and consensus roles.
prepare_anchored_subjects_each	Prepare anchored subjects with one reading moment per row.
AnchoredEstimate	Consensus, moved a fraction of the way toward a forecaster.
ConsensusGap	How far one forecaster stands from the consensus.
AnchoringParameters	A fitted, versioned coefficient set for the anchoring estimate.
assess_eligibility	Assess whether each subject is publishable, and record why.
EligibilityReason	Why a subject is, or is not, publishable.

Bayesian family

Preparation, the posterior, its settings, presets and refusal reasons.

prepare_bayesian_subjects	Read the panel as of one moment and attach every history the posterior needs.
prepare_bayesian_subjects_each	Prepare Bayesian subjects with one reading moment per row.
BayesianPosterior	Prior, model and consensus, blended by their own precision.
BayesianSettings	The arguments that change the posterior.
BayesianPreset	A named, versioned Bayesian configuration with its provenance.
PresetStatus	How far to trust a preset.
PresetProvenance	Where a preset’s values come from.
posterior_preset	Return a preset’s posterior settings, optionally with overrides.
describe_bayesian_presets	Tabulate every preset’s intent and provenance, one row each.
PosteriorReason	Why a subject is, or is not, publishable under the posterior.

Bayesian panel workflow

Separate panel workflow; see its reference contract before switching APIs.

bayesian_kpi_predictions	Apply the Bayesian mathematics to point-in-time panel data.
BayesianPredictionSettings	Versioned settings for the complete Bayesian workflow.
bayesian_preset	Return a preset’s workflow settings, optionally with overrides.

Track-record family

trailing_skill	Attach each row’s trailing skill, computed from its entity’s prior rows.
TrackRecordWeighting	Weight the gap by how well the forecaster’s skill tercile has predicted surprise.

Method protocol and registry

Method	A named post-processing step, fitted before it is applied.
FittedMethod	A method with its parameters settled, ready to apply to any rows.
MethodRequirements	What a method needs before it can run.
available_methods	Return every registered method factory, keyed by name.
create_method	Construct a registered method by name.
register_method	Make a method constructible by name, for configuration files and CLIs.