Quality scores
Quality Scores are measures of data quality calculated at the field, container, and datastore level and recorded as time-series enabling you to track movement over time. A quality score ranges from 0-100 with higher scores indicating higher quality. We'll look at how quality scores are measured for each of those data assets in turn:
Quality Scoring a Field
Contributions to field scores decay over time so that only relatively recent inputs directly influence a field's score. The following inputs contribute to a field's quality score:
- its ratio of valid record anomalies to records
- whether it is checked for expected completeness
- the number other checks defined for it
- its consistency
Quality Scoring a Container
Contributions to container scores decay over time so that only relatively recent inputs directly influence a container's score. The following inputs contribute to a container's quality score:
- the mean of all its fields' scores
- its ratio of valid shape anomalies to records
- whether it is checked for expected volumetrics
- the frequency with which it is scanned
- whether it has a Freshness SLA defined
- the duration of any Freshness SLA violations
Quality Scoring a Datastore
Contributions to datastore scores decay over time so that only relatively recent inputs directly influence a datastore's score. A datastore's quality score is the simple mean of its containers' scores.