Group By monitors
Group By monitors enable you to track data quality metrics across specific segments of your dataset. Instead of monitoring a metric for a column as a whole, you can break it down per category (for example, per region, per school year, per status).
This functionality is especially valuable when:
You want to detect anomalies at a more granular level, within each segment or category.
You need visibility into how data quality differs across categories.
You want to monitor trends and patterns that would otherwise be hidden in aggregated metrics.
Only one Group By monitor can be configured at a time.
Because a Group By monitor spawns multiple monitors (one per category), limiting this to a single configuration helps manage performance.
Explore Group By monitors
When a Group By monitor is active in a dataset, results are displayed at the bottom of the Metric Monitors tab, on the Column Monitors table:
There, you will see:
Entry in Column Monitors table
A Group By monitor is listed like any other monitor, but its description indicates the Group By column(s) and the metric being measured (e.g. "Maximum length of Bus_No grouped by Breakdown_or_Running_Late").
From the Column Monitors table, it is possible to turn on notifications at the column level by clicking on the bell icon. Note that notifications at a category level are not available at the moment.
Groups table
Expanding the monitor displays a groups table, which shows the results for each group or category. Each row corresponds to one category (or combination of categories if multiple columns are grouped). From the groups table, it is possible to delete specific categories by clicking on the bin icon on the right.
Deleting from the groups table is intended to remove groups/categories that are no longer present in the data.
Deleting a category removes the history for that monitor.
If the group/category is still present in the data, the monitor will be re-created on the next scan. It will not be backfilled, unless a historical metric collection scan is triggered.
Example:
Group By Breakdown_or_Running_Late + metric Maximum length of Bus_No → a row for each Breakdown_or_Running_Late value, with the maximum bus number length observed in that category, alongside its anomaly detection status.
Add Group By monitors
You can add a Group By monitor from the Metric Monitors section of the dataset page.
Scroll to the Column Monitors table and click Add Column Monitors.
In the Add Column Monitors panel, toggle on Group By.
Select one or more columns to group by.
For the time being, only columns with a maximum of 50 distinct categories are eligible for Group By monitoring.
(Optional) Exclude specific categories (segments) that you don’t want to monitor.
Select one or more columns to monitor under Column Selection.
Enable one (or more) metric from the right-hand list.
Click Add 1 Monitor on the top right to save.
The monitor now appears in the Column Monitors table and starts tracking anomalies across each category.
Category management
Categories can be excluded when configuring the monitor. See Step 4 on #Add Group By monitors.
Categories can be deleted after creation from the #Groups table if you decide they should no longer be monitored.
Key Considerations
One Group By monitor at a time Only one configuration is allowed, since Group By monitors expand into many underlying monitors.
Multiple Group By columns More than one column can be selected, but the categories generated are combinatory.
Category limits Columns with more than 50 categories cannot be used for Group By monitoring.
Exclusions and deletions You can exclude categories at configuration time or delete them later from the #Groups table.
Notifications Notifications are configured at the column level, not yet at the per-category level.
With Group By monitors, you gain more granular visibility into your data quality, while keeping control over compute cost and category management.
Last updated
Was this helpful?