Using data lineage on a dataset

Audrey M. Updated by Audrey M.

Where to find a dataset's lineage

To see a dataset's lineage, in your back office go to that dataset and click on the Lineage tab.

To review or better understand how data lineage works, see here. If you know more about the lineage dashboard, see here.

Understanding what you see

Note that lineage data is updated after the publication of an ODS object. Only your published datasets can be part of a data lineage.

To make navigating and understanding the graph easier, there's a legend you can access by clicking on the i icon at the lower left corner of your screen.

Objects

You can identify your dataset block, those that belong to you, as well as workspaces that have activated incognito mode.

There are icons for different types of objects, as well as for their status and those of their linked fields.

Relationships

Lines represent relationships between two objects, or else between the fields linked between these two objects, indicated by dotted lines. They are red when a relationship is deemed invalid.

For fields, additional information is provided if the field is used to create a filter or a join key.

Interacting with the diagram

Display Field-level lineage

To make the diagram clearer and more readable, we have chosen to display only the fields that have an impact on other objects or visualizations.

When no field from an origin dataset is specifically used in a federation or source-type relationship, the following message appears: 'All fields are synchronized but not linked.'

The menu LINKED FIELDS displays the fields that are linked to other objects or used in dataset views (analyze, map, calendar and image views).

Datasets views

A subsection Exclusively in its visualizations may appear when fields not belonging to the dataset are used in tooltips or custom views.

By default, the Table view is automatically excluded from data lineage as its functionality remains indifferent to the presence or absence of fields, unlike the map, which is contingent on a geometric field. If a field Geom of type ‘geoshape’ is deleted, the Table view remains unaffected and error-free. However, the map will encounter an error, leading to improper display

Get details about the errors

The elements causing the error are highlighted in red with a red background. Affected fields are marked with an exclamation point (!). Red borders around the objects indicate that the data is no longer available. Here are a few examples:

  • Case 1: Dataset A is federated by Dataset B. If A is deleted, both A & B become unavailable because B is a view of Dataset A (the data is not duplicated).
  • Case 2: Dataset A is joined with Dataset B. If A is deleted, A becomes unavailable but B remains available. The publication will fail because A has been deleted.

A tooltip is displayed when hovering over an invalid relationship. It provides information about the root causes of the error.

Highlighting objects

When you click on a button of the legend or on the top of a card, only the linked objects and direct relationships are highlighted. This helps you focus on the information you need.

How similar objects are grouped

The lineage for datasets that have relationships with a large number of ODS objects would quickly become unreadable if every object was always depicted. To make things easier to understand, any time there are four or more similar objects, they're grouped according to a few simple rules:

  • They are the same type of ODS object (dataset, page, etc.)
  • They are related to a common ODS object, and have the same type of direct relationship
  • They have no downstream relationships
To avoid overloading the diagram, third-party objects are systematically grouped whenever a dataset has more than one source.

Hover to expand the relationship label

You can also hover over the abbreviated relationship indicator on the diagram and it will expand, so here, the "F" expands to "Federated by."

Access ODS objects directly

You can click any ODS object title or identifier (colored block) to be redirected to the object page to check its details (explore view).

Expand the view

You can switch the view to full screen, and navigate or zoom in or out as needed. Des contrôles sur la gestion des regroupements et l'affichage des titres sont disponibles.

How did we do?

Data lineage: what it is and how it works

Using the data lineage dashboard

Table of Contents

Contact

Powered by HelpDocs (opens in a new tab)