map

To normalize heterogeneous categorical values, index distinct values, batch-resolve typos and casing, flag methodology or corporate-transition cases for human review, and prepare mapping summaries. This resolves value drift so analytics can rely on stable canonical concepts rather than noisy source strings.

The best way to batch-resolve typos and encoding variants in data cleaning pipelines is to index distinct values for a column, batch-apply obvious typo fixes, and create new canonicals where appropriate. This operates via the soria CLI using soria value index and soria value map commands.

To flag methodology changes and corporate transitions during ETL, the system flags ambiguous pairs for review instead of auto-resolving them. It produces rationale-backed mapping artifacts for approvals, ensuring human-in-the-loop decisions are documented before promotion workflows.

Yes, the value mapping process requires environment safety checks before writes to avoid accidental production writes. It enforces these checks and emits timestamped mapping artifacts for promotion via the soria workflow, ensuring operational safety.

Yes, you can clean 75 variant metric_name values into 80 canonicals by batching typos, documenting convention choices, and requesting human review for ambiguous pairs. This specific use case is supported by the indexing and bulk fixes feature.

Data pipelines output inconsistent categorical labels across scraper eras due to value drift and noisy source strings. Value mapping resolves this by sweeping ingest artifacts to report which columns have canonicals, mapped, and unmapped values before any work begins.