curator

Recipe-driven creation of canonical entity datasets.

Curation starts from a harmonized entity, incorporates enrichment evidence, and applies registered reconciliation, imputation, inference, and filtering steps before saving a canonical entity dataset.

Submodules

• diagnostics
• evidence
• filters
• formatters
• imputers
• indicators
• inferers
• occupancy
• provenance
• reconcilers

Classes

CurateState	Mutable container passed through every curation pipeline step.
Curator	Create a canonical entity dataset from harmonized and enriched inputs.

Functions

curate(→ None)

Instantiate and run curation for recipe.

Package Contents

class openplaces.io.curator.CurateState

Mutable container passed through every curation pipeline step.

class openplaces.io.curator.Curator(recipe: str | dict, admin_ids: str | list | None = None, verbose: bool = False, save_statistics: bool = False, skip_steps: str | list | set | None = None)

Create a canonical entity dataset from harmonized and enriched inputs.

curate(reprocess: bool = False, cleanup: str | None = None) → None

Run curation for all configured administrative units.

cleanup='consumed' deletes this recipe’s direct inputs after each admin unit finishes, iff every consumer in the recipe tree is complete (see cleanup_consumed_inputs()).

show_random_entity()

Plot a random entity from the first configured admin unit.

Delegates to openplaces.viz.maps.show_random_entity().

openplaces.io.curator.curate(recipe: str | dict, admin_ids: str | list | None = None, reprocess: bool = False, verbose: bool = False, save_statistics: bool = False, skip_steps: str | list | set | None = None, cleanup: str | None = None) → None

Instantiate and run curation for recipe.

When save_statistics is True (or the recipe sets save_statistics: true), curation steps write diagnostic tables to the cache (e.g. geometry-indicator quantiles and use-group separability) without changing the curated output.

Pass skip_steps (a step name or a collection of names) to skip computation-intensive pipeline steps for this run without editing the recipe; a recipe step may also be disabled persistently with enabled: false.

cleanup='consumed' deletes the recipe’s direct inputs after each admin unit finishes, iff every consumer in the recipe tree is complete.