> For the complete documentation index, see [llms.txt](https://cleyrop.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://cleyrop.gitbook.io/docs/v4.7-en/data-and-ai-project/datasets/modifier-le-schema-dun-dataset.md).

# Modify a dataset's schema

Rename, convert, or delete columns from an existing dataset from its details page, without transformation.

The owner of a dataset can evolve its schema directly from the dataset page, without going through a transformation: rename a column, widen its type, delete it, or complete its description.

***

## Edit columns

1. Open the tab **Schema** tab on the dataset details page.
2. Click on **Edit**.
3. For each affected column:
   * **Rename** : enter a **New name** (lowercase letters, numbers and underscores, not starting with an underscore).
   * **Change type** : choose a type from the list. Only conversions **without data loss** are offered (see below). A Text column cannot change type.
   * **Delete** : the column is greyed out; click on **Restore** to undo as long as you haven't saved.
   * Edit the **description**.
4. Click on **Save** : all changes are applied at once.

{% hint style="warning" %}

* Deleting a column is **irreversible** after saving, and a dataset must keep at least two columns.
* For a dataset from a Dataflow, columns present in the deployed version cannot be renamed, converted, or deleted.
* The change is applied asynchronously: if a schema change is still being synchronized, try again later.
  {% endhint %}

## Allowed type conversions

Only conversions that cannot alter existing data are offered:

<table><thead><tr><th width="260">Current type</th><th>Possible types</th></tr></thead><tbody><tr><td>Integer (Integer)</td><td>Long integer, Decimal (Double), Text</td></tr><tr><td>Long integer (Long)</td><td>Decimal (Double), Text</td></tr><tr><td>Decimal (Float)</td><td>Decimal (Double), Text</td></tr><tr><td>Decimal (Double), Exact decimal, Boolean</td><td>Text</td></tr><tr><td>Date</td><td>Timestamp, Timestamp with time zone, Text</td></tr><tr><td>Timestamp, Timestamp with time zone</td><td>Text</td></tr><tr><td>Text, Binary, Array, Structure, Key/value pairs</td><td>No conversion</td></tr></tbody></table>

## Effect on subsequent imports

After renaming or deleting a column, a new import of the original file depends on the [schema policy](/docs/v4.7-en/data-and-ai-project/datasets/politique-de-schema.md) of the dataset: with **Schema merge**, the column is recreated from the file header; with a **strict fixed schema**, the import is rejected. Update your source files after a schema change.

***

## Best practices

* Warn users of Dataflows, Agents and exports that consume the dataset before renaming or deleting a column.
* Prefer widening a type (Integer → Long integer) rather than converting it to Text, which loses possible calculations on the column.
* Complete the **description** of the columns at the same time: it is visible on the dataset page and via the API.