Create alerts based on filter conditions applied to table fields. You can add single or multiple conditions to observe specific data subsets. Use filtered datasets to generate volume, data drift, and freshness alerts for the exact data you need to monitor.
Profiling statistics are calculated from the filtered dataset only, not the entire table. For example, if a table has 100 total records but your filter returns 10 records, all statistics are based on those 10 filtered records.
- Go to and select + Create Observer.
- Select the datasets and fields you want to observe, then click Next.
- Select the Filter Selected Datasets link.
- Select the dataset to filter.
- Select the field, operator, and value. Add multiple
conditions as needed. Choose Match
All for logical
ANDor Match Any for logicalOR. - Complete the remaining Observer creation steps.
The Observer saves with your filter conditions, and you're redirected to the Observers page.
- Datetime values are assumed to be in UTC timezone.
- Float values are rounded to two decimal places for filtering. For example, 100.2367 becomes 100.24.
- The where clause can reference columns not in the observed column list.
- Columns in the where clause must be cataloged.
- Changing filter conditions is treated as a data change, which may increase or decrease volume and trigger alerts.
Limitation with quoted database objects: Tables or columns with single or double quotes in their names (for example, 'TableName' or "ColumnName") can cause Observer runs to fail. Rename affected objects without quotes.