Re-Running a Dependency Analysis

You can re-run a dependency analysis on an entity. Re-run the analysis in the following situations:

■  To verify dependency information

■  To change parameters or options to test for different results

If suitable dependencies were not found during the data import or expected attributes were not identified as dependencies, re-run the dependency analysis to do the following:

■  Find potential dependencies that are less than 98% consistent.

■  Find dependencies with more than two combined attributes.

■  Include attributes that are less than 2% distinct.

► To re-run a dependency analysis

1.

Do one of the following:

■

On the Home tab, select Keys/Deps > Discover.

■

In the Navigation View, right-click an entity and select Discover Keys or Dependencies.

■

In the Navigation View, from the Analysis tab, right-click Dependencies and select Discover Dependencies.

The Discover Keys or Dependencies window opens.

2.

Click the entity you want to re-analyze.

3.

Enter a Job Name.

4.

Select Find Dependencies.

5.

Click Next.

6.

Use the Sample options to specify which data rows you want re-analyzed.

Option	Description
How many rows do you want to sample?	Enter a number that indicates the total number of rows you want analyzed. For example, 1000.
Start sampling at row	Enter the row number at which you want to start counting the number of rows to be analyzed. For example, 1 indicates to start at the first row.
Which rows do you want to include in the sample?	Select one of the following options to indicate which rows to include in the analysis: ■ Every Row - Specifies to include all rows in the sample. ■ Every n Row - Enter a number to indicate how many rows to skip before selecting the next row to include. ■ Random n % - Enter a number to indicate the total percentage of rows to include from the sample. Note: The random % is the percentage chance of rows being included. Therefore, the actual number of rows loaded may be different for each load of the same file even if you specify the same percentage.
Re-use previous Random selection	Check this box to re-analyze the same set of data as the previous random analysis. This box is grayed if there is no previous random analysis. Example If the last random analysis sampled from rows 1, 3, and 5, then checking this box performs an analysis for the same rows.

7.

(Optional) Click Defaults to reset the options and start again.

8.

Click Next.

9.

In Dependency Analysis, specify the parameters for running the analysis.

Option	Action
How consistent should your dependencies be? At least n %	Enter a percentage that indicates the level of consistency that qualifies a dependency. Example Entering 98 indicates that you want to find trends in the data that are consistent 98% of the time. Take the example of two attributes "Gender Code" and "Gender Description". If Gender Code has a value of "1", and Gender Description has a value of "Female" in at least 98% of the rows, this will be flagged as a potential dependency.
Which attributes do you want to exclude? Those less than n % unique.	Enter a percentage that indicates the percentage of uniqueness required to include an attribute. Example If you entered 2%, any attribute that is less than 2% distinct is excluded.
What type of dependency do you want to discover?	Click one of these options: ■ Single Attribute - Dependencies based on one left-hand side (LHS) attribute ■ Multiple Attributes - Dependencies based on combined left-hand side (LHS) attributes
Maximum number of attributes that can form a compound?	If you select Multiple Attributes, enter the maximum number of combined attributes to include in the analysis.

10.

(Optional) Click Defaults to reset the options and start again.

11.

Click Finish.

12.

Run the analysis now or schedule a time for the job to run later.

Note:

To monitor the progress of an analysis, on the Home tab click Background Tasks. Look for an Activity Name with the type Key & Dependency Analysis and your user name under Created By.

Related Topics

---