Skip to content

Predicted By Anomaly Reporting

How the Predicted By check reports violations: the anomaly messages it produces, what the numbers mean, how the offending record appears in Source Records, and how to replace the default message with a value from your own data.

Anomaly Messages

At 100% coverage, Predicted By reports violating rows as Record Anomalies. Below 100% coverage, a failed coverage assertion produces one Shape Anomaly for the dataset. The two templates are:

Record Anomaly

The field '<field_name>' has value '<value>', which is not within the predicted range defined by <expression> +/- <tolerance>

Shape Anomaly

For the field '<field_name>', X.XXX% of N records (K) are not within the predicted range defined by <expression> +/- <tolerance>

When a filter is set, both Record and Shape Anomaly messages end with [filter: <expression>].

Scan settings can group a large number of Record Anomalies into one rolled-up Shape Anomaly. This behavior applies across rule types and is documented under Maximum Record Anomalies per Check.

What the Numbers Mean

  • X.XXX%: the fraction of filtered rows whose value falls outside the predicted range.
  • N: the total number of rows the check evaluated (after the filter, if any).
  • K: the number of rows outside the band, including any where the expression returned nothing.

Source Records Behavior

The offending cell is highlighted with an orange outline and an orange-tinted background in the Source Records view. The message carries the expression and the tolerance, so read the highlighted value against the other columns on the row to see what the prediction would have produced.

Custom Anomaly Description

Predicted By supports Custom Anomaly Description because it emits Record Anomalies. When anomaly_message_field (or the Custom Anomaly Description toggle in the UI) is set to another column on the same row, the Record Anomaly message becomes the value of that column for the violating row. When the referenced column is null, missing, or empty, the standard template is used instead.

Because the option only applies to Record Anomalies, it does not affect Shape Anomalies, which always use the fixed template.

See Also

  • Examples

    Three production scenarios with sample data, anomaly messages, and the SQL equivalent of what the check evaluates.

    Examples

  • Best Practices

    Guidelines for writing the prediction, sizing the tolerance, and choosing between this rule and a simpler one.

    Best Practices

  • Permissions

    The team permission each action needs: view, create, edit, archive, restore, and delete.

    Permissions

  • How It Works

    The complete reference: definition, field scope, the prediction expression, how the tolerance band works, date and time targets, NULL handling, filter behavior, and coverage.

    How It Works