[In preview] Improvements to managing freshness alerts

Following up on our recent improvements to Volume alerts, we've released a similar set of improvements for managing Freshness alerts (time since last update & time since last row count change). Same as the Volume release, we’re rolling this out gradually over the next few weeks. It is currently available to just a subset of customers.

Specific improvements:

• Freshness thresholds don't automatically widen after an anomaly. Instead, anomalies are excluded by default from the set of training data. If they don’t want to be alerted to similar anomalies in the future, users can ‘Mark as normal’. This will re-introduce the anomalous data point to the training set and widen the threshold.
• Users can 'select training data' directly in the freshness chart. This gives the user full control over which data is used to train the model, without needing to navigate to Settings to create Exclusion Windows.

With this change, all anomaly detection in Table monitors (Freshness & Volume) now behaves the same way. Anomalies are excluded anomalies by default from the data set that trains thresholds. And users can then choose to add them back in if they’d like to widen the threshold.

Read more about interacting with our anomaly detection.