> For the complete documentation index, see [llms.txt](https://docs.decube.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.decube.io/data-quality/monitor-configuration-settings/retraining-monitors.md).

# Retraining Monitors

When you modify certain settings on an existing monitor, Decube discards the monitor's historical data and retrains the ML model from scratch. Understanding which changes trigger this — and what gets deleted — helps you avoid unintended data loss.

{% hint style="danger" %}
A retrain deletes all historical test results, metrics, cases, and incident history for the monitor, including any **open incidents**. Note open incidents before reconfiguring a monitor if you need to preserve that information.
{% endhint %}

## What triggers a retrain

The following configuration changes trigger an automatic retrain:

| Setting changed                                              | Notes                                                                                                                            |
| ------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------- |
| Row creation mode (Timestamp / SQL Expression / All Records) | Changing mode resets the data collection baseline                                                                                |
| Timestamp column                                             | Changing the column changes what the model is trained on                                                                         |
| SQL Expression value                                         | Any change to the expression text                                                                                                |
| Smart Training toggled on                                    | Enabling Smart Training after creation starts a fresh training run                                                               |
| Scan frequency                                               | Changing frequency changes the lookback window (see [How Anomaly Detection Works](/data-quality/anomaly-detection-explained.md)) |
| Timezone                                                     | Changes the alignment of scan windows                                                                                            |
| Time of day                                                  | Changes when scans run                                                                                                           |
| Day of week                                                  | Applies to weekly-frequency monitors                                                                                             |
| Day of month                                                 | Applies to monthly-frequency monitors                                                                                            |

## What gets deleted during a retrain

When a retrain is triggered, Decube permanently deletes the following for that monitor:

* All historical test results and metric values
* All cases associated with past scan runs
* All incident history
* All **open incidents** — these are closed and deleted, not resolved

There is no recovery path for deleted incident history. If you need to retain a record of open incidents, resolve or export them before making any of the changes listed above.

## What does not trigger a retrain

The following changes are safe to make without triggering a retrain or losing any data:

| Setting                                                            | Safe to change |
| ------------------------------------------------------------------ | -------------- |
| Monitor name                                                       | Yes            |
| Monitor description                                                | Yes            |
| Notification channels                                              | Yes            |
| Incident level                                                     | Yes            |
| Threshold bounds (Absolute, Percentage, Positive Range, Any Range) | Yes            |

## Operational guidance

Before reconfiguring a monitor that uses Smart Training:

1. Go to **All Monitors** and filter to the monitor you plan to change.
2. Review any open incidents and take note of them, or resolve them if they are addressed.
3. Make your configuration change. Decube begins retraining automatically.
4. The monitor enters a training period during which no new incidents are generated. Check the [training duration](/data-quality/anomaly-detection-explained.md#historical-lookback-by-scan-frequency) for your scan frequency to know when to expect incidents to resume.

{% hint style="info" %}
If you need to change a setting that triggers a retrain on a high-volume production table, consider the timing carefully. The monitor will be inactive during the training period, which can range from hours to days depending on the scan frequency.
{% endhint %}

## Settings that cannot be changed after creation

Some settings are locked after a monitor is created. To change them, you must delete the monitor and create a new one:

* **Asset** (source, schema, dataset)
* **Test type** (for example, you cannot change a Freshness monitor into a Volume monitor)
* **Monitor mode** (Scheduled vs On-Demand)
* **Group by** configuration


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.decube.io/data-quality/monitor-configuration-settings/retraining-monitors.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.