The README discusses what is covered in this repo.
- clear delineation between "anomaly detection" and "broken" data (e.g., broken expectations/contracts/promises)
- Anomalous data might be a good or bad surprise (more revenue than expected 👍 or more complaints than ever before 👎) but the data might still be correct and not "broken"
- Broken data violates some fundamental expectation of the data, like no nulls or unique identifiers
- Configuing
error
vs.warn
severity
levels in dbt
- the new pytest-based testing framework (that can be used for adapters and packages alike)
- or alternatives like pytest-dbt-core
- unit testing:
- dbt_unittest
- dbt-datamocktool
- dbt-unit-test
- any other way of "unit testing" dbt transformations
- test coverage:
- structural conformance testing:
- data comparisons
- other stuff related to data observability: