Wikis & Flow Documentation

Wikis are a widely-used tool for collaboration, allowing team members to create and edit articles that document a project, a group of projects, or more. Learn about wikis and other tools for documenting projects in this section.

Concepts

• Concept | Workflow documentation

Reference | Editing wiki articles

Dataiku supports Markdown and HTML/CSS for formatting wiki articles. In Edit mode, the controls along the top of the edit area and the sample text should get us through common formatting options, and we can consult the reference documentation for details on Dataiku markdown support.

One feature worth calling out specially is the ability to reference Dataiku objects within the wiki. You need to provide the name of the project that contains the object, the type of object, and then the name of the object itself. This creates a link in the finished wiki that teammates can follow.

The Actions menu allows you to:

• Set home article. This is the article displayed first when you visit the wiki.

• Change parent article. This changes the hierarchy of the wiki to place this article at the top level of the wiki or as the child of another article.

• Switch to article/folder layout. “Article” layout displays the article text with a list of attachments in the right panel. “Folder” layout displays the list of attachments below the article text, and the article can be expanded/collapsed for reading.

• Rename. Rename the article at any time. Take care, though, since this will break existing links to the article.

Reference | Sharing and promoting wikis

Your wiki is now visible to teammates browsing the project, and there are a few ways to broaden your audience.

Dashboard insight

You can add articles as insights on a Dataiku dashboard. This is useful for single-sourcing project reports, which can be written in the wiki for teammates, and then shared on the dashboard for readers.

Promoted wikis

You can make a project wiki easier to find from the Dataiku homepage by promoting it. From the project’s Settings > Wiki, select Display the wiki of this project in the global (instance-wide) wikis list. The wiki will then appear in the list of wikis displayed on the Dataiku homepage.

Home articles

The Dataiku administrator can select wiki pages to be used as “home articles” to convey the organization, procedures, and guidelines for using the instance and projects on the Dataiku instance.

• From Administration Settings > Themes & Customization, select the articles to add.

Home articles appear as tabs on the “Getting Started with Dataiku” You can see the homepage articles that are based on wikis in projects you have access to.

How-to | Create a wiki article

Every project in Dataiku is wiki-enabled. To create your first article:

• Go to the Wiki section of the top navigation bar and click +Create Your First Article

• In the New article dialog, give the article a name. For the moment, we can call it README to signal that this is the wiki article team members should start reading.

• Since this is our first article, choosing the Sample wiki article template will provide us with a helpful outline and markdown examples.

• Click Create.

How-to | Export a wiki to a PDF

Your data team has documented a project using wikis in Dataiku, but now you need to share that knowledge with stakeholders who do not have access to the project. Learn how to export a wiki from Dataiku to PDF.

You will need a Dataiku instance set up to export to PDF.

From the Actions menu of a wiki, select Export to PDF. You’re given options surrounding what to export (which articles should be part of the PDF, and whether to include wiki attachments in a zip archive), and format (page size of the PDF).

If you need to create regular PDF exports of this wiki, there are two ways to do this in scenarios.

One way is to create a scenario step that exports the PDF to a managed folder. The PDF is then directly accessible to anyone with permissions on the folder, or can be redistributed through further code.

Another way is to attach the PDF export to a scenario reporter in the scenario settings.

For further details, see the reference documentation on wikis.

How-to | Export Flow documentation with the Flow document generator

The Flow Document Generator creates a snapshot of the elements in your Flow, including a project description, an image of the Flow, and details of each dataset, recipe, and folder.

Flow documentation helps project teams make their data processes more transparent and explainable to others for AI governance and collaboration purposes.

Export the documentation

To export Flow documentation for your project:

• From the Flow, click on Flow Actions in the bottom right.

• Choose Export documentation.

• You can upload a custom template, but for now choose Use the default template and Export.

• After the Flow documentation is ready, choose Download.

• Open the downloaded document, which is formatted for Microsoft Word, and refresh the table of contents to view the complete documentation.

Take a few minutes to read through the document, which includes:

• the project description

• an image of the Flow

• detailed information about each dataset, such as the connection type, file size, original creator, schema and related recipes

• related recipes for datasets

• and other implementation information.

Note

To create a template customized to your company’s needs, you can edit the format and variable references in a Microsoft Word file, and supply your own custom template to the Flow Document Generator.

How-to | Automate Flow documentation exports

You can add a Flow documentation export as a scenario step in your machine learning model so Dataiku will automatically generate it and add it to the model documentation.

Before adding this step, make sure your project includes a folder to save the Flow documentation. Folders appear in the Flow with a folder icon. To add a folder:

• Navigate to the Flow and choose +Dataset > Folder.

• Give the folder a Label and select Create.

With a folder created, you can now add the Flow documentation step to a scenario.

• From the Flow, select the Jobs menu from the top navigation bar.

• Choose Scenarios.

You can add a step to an existing scenario, or create a new scenario if your project does not have any existing ones.

• Choose +New Scenario in the top right and give the scenario a name then choose Create.

• Within the scenario, navigate to the Steps tab.

• Select Add Step in the bottom left, and then select Export flow documentation.

• In Destination folder choose a folder where you want Dataiku to save the Flow documentation.

The Flow documentation will now automatically export when you run the scenario and be saved in the designated folder in the Flow.

• Select Run in the top right.

• After the job completes, navigate to the Flow and double-click on the folder you designated to hold the documentation.

A new version of the documentation will appear in the folder each time you run the scenario. In the folder, you can view the date and time each document was last modified, as well as download, delete, rename, move, or publish the documentation.

Note

You can also add steps to export model documentation, code notebooks, wikis and other documentation in the scenarios.