Reviewing Dependency Analysis Results - trillium_discovery - trillium_quality - 17.1

After your data is imported into a repository and you have created an entity, you should review the potential dependencies and determine whether they are relevant to your data investigation.

To review potential dependencies in an entity

1. In the Navigation View, click the entity in which you are interested.

   The Summary Data View for the selected entity opens. (If the view does not open, click the Metadata Summaries icon.)

2. Click the Relationship Summary tab.
3. Under Dependencies, note the number of discovered dependencies.
4. Click Discovered.

   The dependencies discovered during the data analysis process display in the List View.

5. Review the information for each dependency.

   Tell me about the dependency information…

   Column Label	Description
   Lh Attrs	A comma-separated list of attributes that form the left-hand side identifier for the dependency. The left-hand side identifier can also be a single attribute.
   Rh Attr	The attribute that is identified by the left-hand side attributes.
   Status	Indicates whether the dependency has been reviewed or not. Discovered means not reviewed Permanent means valid dependency
   Verified	Indicates whether the dependency was checked against every row of data, and therefore, whether the list of contradictions is complete. If the original data source contains less than 10,000 rows, then this column has a value of YES because all rows were included in the initial dependency search process. The data import process samples a maximum of 10,000 during the initial dependency analysis.
   Job	The schedule job number that corresponds to the column named Ref associated with dependency-related activity in the Background Tasks list.
   Quality %	The measure, as a percentage, of how consistently a left-hand attribute (Lh Attr) identifies the value in the right-hand attribute (Rh Attr). Dependencies with Quality % equal to 100 have no conflicts. Dependencies with less than 100% contain conflicts. Example If the dependency between ORDER_NUMBER and ORDER_DATE has a Quality % equal to 99%, this indicates there is at least one row with a conflicting value, but that 99% of the values adhere to the dependency rule.
   Confirming LR Values	The number of values on the left-hand side (LHS) of the dependency that predict right-hand side (RHS) values.
   Conflicting LH Values	The number of values on the left-hand side (LHS) of the dependency that do not determine the right-hand side (RHS) values.
   Resolved %	Indicates the status, as a percentage, of the left-handed conflicting values that have been resolved, if any.
   Conflicting Rows	Total number of rows that contain a conflicting value for the right-hand side (RHS) attribute of the dependency.
   Verified Date	Date on which the dependency was last verified.
   Verified By	Name of user who verified the dependency as valid.
   Normalise Flag	Indicates whether the value can be included in a normalization task. 1 equals True/Yes, 0 equals False/No.
   Created Date	Date on which the dependency was created.
   Created By	Name of the user who created the dependency.
   Dependency Key	Indicates the reference numbers of the attributes used in the dependency.

6. Determine if a dependency is relevant by double-clicking a dependency with a Quality % value that is less than 100%.

   The Dependency Conflicts List View opens.

   Note: You can usually identify which dependencies are relevant to your project from the Dependency Conflicts List View. To explore the reasons why a conflict may exist, you must drill down to show the actual rows where the conflicts exist.