When running a glue data quality evaluation, how can I filter the data it is run on?

Question

I'd like to run evaluations of my data quality rulesets on single partitions of my table rather than the whole table. This is because for most of my tables each partition effectively represents a snapshot of the data and running the checks only makes sense in the context of a single partition. Is there a way to filter or subset the data that a ruleset is evaluated on?

Preferably I'd like to do this when triggering the evaluation, but defining a restriction in the DQDL rules might also work.

Answer

The `start_data_quality_ruleset_evaluation_run` API allows you to pass a `pushDownPredicate` in its Additional options section. You can also similarly pass a CatalogPartititionPredicate

```
aws glue start-data-quality-ruleset-evaluation-run --ruleset-names ruleset1 --region us-east-2 --data-source '{"GlueTable":{"DatabaseName": "db", "TableName":"tb", "AdditionalOptions": {"pushDownPredicate": "year=\"2022\""}}}' --role Admin
```
A similar API structure can be use for getting recommendations post filtering as well

When running a glue data quality evaluation, how can I filter the data it is run on?

Relevant content