A dependency is a many-to-one data relationship. In the Discovery Center, you can create expected (permanent) dependencies that you know are in your data. You can also discover potential (possible) dependencies, review the results, save them as permanent or delete them if they are not valid, then add another discovered dependency with a different sample size, uniqueness threshold, and number of attributes.
Use the Create or Discover Dependencies window to guide you through the Select, Configure, and Finish tabs, which include the following tasks (depending on the type of dependency you are adding):
- Select whether to create or discover the dependency:
- Creating a dependency adds a new, permanent dependency to the repository.
- Discovering a dependency runs analysis on a sample of your data to find potential dependencies.
- Select left-hand attributes and a right-hand attribute.
- Configure dependency analysis options.
- Configure the sampling options.
- Add dependencies for profiled (fully-loaded) data sources only. Dynamic data sources are not supported.
- Dependencies are added to the repository to which you are currently logged in.
- After you add a dependency, you can change its status at any time. For example, when you add a potential (discovered) dependency you can then change its status to permanent after you validate that the relationship is useful.
- After you add dependencies, you can drill-down to see the LHS and RHS attributes, confirming values, and conflicting values and rows.
- Create a dependency when an expected dependency was not found during data load or when you discovered a dependency.
- The Discovery Center automatically performs dependency analysis on a sample of your data (up to 10,000 rows) during data load (data source creation) to discover possible (potential) dependencies. To improve load performance, administrators may disable dependency analysis by clearing the Discover Dependencies at Load Time selection when editing loader connections in the Repository Manager. When dependency analysis is not run during data load, follow this procedure to run analysis.
- When you create a dependency, the process does not sample the data to validate your dependencies; the full data volume of the data source is evaluated.
Using the Create or Discover Dependencies Window
The Create or Discover Dependencies window guides you through the process of adding a permanent or discovered dependency for the current data source. Your work is saved as you continue through the tabs. At any time during the process you can view a summary of your selections and make edits to your work. Navigate to other areas of the Discovery Center as needed while the window is open.
To open and use the Create or Discover Dependencies window-
On the Summary tab, in the Relationships table, click the Dependencies
plus button (
). The Create or Discover Dependencies window opens
with the Select tab active.
-
To edit your selections:
- Click the Configure or Finish tab when it shows a green check.
- Click Summary at any time to see an overview of your saved selections. From the Summary page:
- Click the edit button (
) for the step you want to edit. This closes the
Summary and opens the step for editing. - Click Back or In Progress to open the last step you worked on.
Clicking the Finish button adds the dependency to the current repository. To close the window at any time without saving your work, click Cancel or the X icon.