Dataset data

Query stored rows and upload files on a dataset's Data tab. For a regular delta dataset, you can also fill in sample data that matches the current schema.

Query data

1. Select the dataset in the collection tree and open the Data tab.
2. Review rows in the table, or switch to Map view or Chart view.
3. If needed, run a SELECT query in the SQL scratchpad.
4. To reload the latest data, select Refresh.

The table shows 25 rows per page. Map view appears when there are valid latitude and longitude columns, and Chart view appears when there are date/time and numeric columns. SQL can query only the current dataset.

Upload files to an existing dataset

Use File upload on delta and object datasets.

1. Select Data → File upload.
2. Choose a CSV or Parquet file. An object dataset stores files as objects and does not check the schema.
3. In Review file upload, check the differences between the current schema and the file schema.
4. Select the enabled upload button.
5. Check whether the upload has completed in the upload progress panel.

Depending on the review result, the button reads Initialize schema and upload, Merge schema and upload, or Upload.

Review result	Handling
New columns found only in the file	Added to the schema during upload.
Temporary field1 schema or empty schema	The schema is initialized from the file columns.
Existing columns missing from the file	Upload is blocked.
Type conflicts or conflicts between files	Upload is blocked.

If upload is blocked, revise the file or edit the dataset schema first. You cannot edit headers or generate column names with AI in this review dialog.

To create a new dataset directly from a file, use Quick Add.

Fill sample data

On a regular delta dataset, open Data → Fill sample data. This action is not shown for object, kafka, or rest datasets or for a delta dataset connected to PostgreSQL change ingestion.

1. Enter the Number of rows to generate, from 1 to 100,000.
2. Select Append or Overwrite.
3. To reproduce the same result, enter a Seed.
4. In Preview, check 20 rows before saving.
5. Select Fill, then check the data and the new version.

If you leave the seed empty, the server also uses the seed it generated for the preview when it fills the data. With a temporary schema that has no actual fields, you must define the schema first.

Overwrite

Overwrite replaces all existing rows. Review the target in the confirmation dialog before continuing. If you need the earlier data, restore a data version under Version History.

Review data versions

In Data versions on the Version History tab, preview a sample of up to 25 rows for each commit. You can also download it as CSV or Parquet.

Metadata changes is the history of the alias, description, tags, options, and schema. Restoring metadata does not change table data.

For how restoration differs by asset, see Version history.