Migrate a dataset to a data model (Beta)

You can migrate a dataset to a data model to take advantage of the capabilities that exist only in data models, such as the following:

• Version tagging and materialization of tagged versions
• Relationships to perform joins only as needed
• Entity relationship diagram of data model contents and relationships

What is migrated from a dataset to a data model

When you migrate a dataset to a data model, Sigma creates a new data model with elements that correspond to the components of the published dataset. Any draft changes and warehouse views are not migrated. The following components of a dataset are automatically recreated in the data model:

🚧

If a materialization schedule is migrated, the materialization runs immediately when the data model is created.

How CLS rules are migrated

Column-level security (CLS) rules and column visibility rules are defined differently in a data model than a dataset. After migration, rules defined in a dataset are set up according to the following table:

For more details about CLS in data models, see Configure column-level security.

How links are migrated to relationships

When you migrate a dataset with links to other datasets, all linked datasets are migrated to one data model. Each dataset is created as a table in the data model. If you choose to update references, all references to the datasets are updated in workbooks that use the datasets as sources.

📘

If you migrate a dataset with links to other datasets, but one of the links references the migrated dataset, the migration cannot be performed. Remove the link that references the dataset to perform the migration, or create a data model and migrate the dataset configuration manually.

For example, if you have a dataset with no links, the dataset is migrated to a new data model. The migration looks like the following:

graph LR
  A[Migrated Dataset] --> B[New Data model]

  style A fill:#FF9A74
  style B fill:#4CEC8C

If you migrate a dataset with links to two other datasets, the dataset and its linked datasets are migrated to one data model. The migration looks like the following:

graph TD
  A[Migrated Dataset] --> B[New Data model]
  A -->|Link| C[Dataset]
  A -->|Link| D[Dataset]
  C --> B
  D --> B

  style A fill:#FF9A74
  style B fill:#4CEC8C
  style C fill:#FF9A74
  style D fill:#FF9A74

How references are updated

When you migrate a dataset to a data model, any documents that use the dataset or linked datasets as a source (reference the dataset) can be updated to instead reference the newly created data model and related columns. When you migrate a dataset to a data model, you can choose whether to update references or not.

If you choose to update references from the dataset when migrating to a data model, the following references are not updated:

• Explorations that use the dataset.
• An unpublished draft of workbooks that use the dataset, if the published version of the workbook does not use the dataset.

Requirements

• You must be the dataset owner or be granted Can Edit access to the dataset
• You must have the Create, edit, and publish datasets permission enabled on your account type.

Migrate a dataset to a data model

To migrate a dataset to a data model:

1. Open the dataset.

2. Select the caret () to open the document menu, and then select Migrate to data model.

3. In the Migrate to data model modal, review the details of what will be created when you migrate the dataset:

   • If your dataset is linked to other datasets, the linked datasets are listed.
   • If your dataset is used by other documents as a data source, the total number of references to be updated is listed.

4. (Optional) If your dataset is used by other datasets, workbooks, or data models as a data source, Sigma can update the data sources of affected elements to use the new data model automatically. Select the checkbox for Update references from dataset to new data model.

5. Select Create.

   Sigma creates a data model in the same location and with the same name as the migrated dataset, with you as the owner. Any linked datasets are created as tables and named to match the relevant datasets.

   The created data model opens in the same browser window. Review the details to ensure that complex configurations migrated properly.

💡

After you migrate a dataset, you cannot migrate it again unless you delete the data model created when you first migrated the dataset.

Review migration status for all datasets in your organization

If you are assigned the Admin account type, you can review the datasets in your organization and the migration status for each one. Optionally migrate datasets from the list, or open a dataset and migrate the dataset from the document menu.

📘

You must be assigned the Admin account type to review the migration status for all datasets in your organization.

1. Go to Home and select the user icon with your initials.
2. Select Administration, then choose Dataset migration.

On the dataset migration page, you can perform different tasks:

• Review the list of datasets in your organization. The following information is available:

  • Dataset lists the name of the dataset with a link to open the dataset.
  • Migration status lists whether the dataset has been migrated. Datasets marked Not required are not used as a data source (are not referenced by other documents) and likely does not need to be migrated to data models.
  • Migrated by lists the username of the user that selected the migration.
  • Migration date lists the time at which the migration was performed.
  • Remaining references lists the total number of references to the dataset by other documents, such as a workbook, data model, or another dataset.
  • Data model lists the data model created by the dataset migration.

• As you review the list, you can search, sort, and filter the list depending on your use case:

  • Search by dataset name to identify a particular dataset.
  • Filter the datasets by Migration status to identify the datasets that still require migration.
  • Sort the datasets by Remaining references to identify the most-used datasets.

• Optionally perform other management tasks for datasets:

  • Select More () > Delete dataset to delete a dataset.
  • Select More () > Migrate dataset to migrate a dataset that has not yet been migrated.